How a Workplace Hazard Becomes a Citation a Company Must Answer

OSHA standards and enforcement decide what happens between the moment someone notices a danger at work and the moment a company answers for it. A safety professional who wants to fix a hazard, an employment lawyer advising a client after an inspection, and a union representative preparing to walk an inspection with a compliance officer all need the same thing: a working map of how a workplace hazard becomes a legally enforceable standard and then a citation a company must answer. This article draws that map. It traces the five mechanisms Congress built into the Occupational Safety and Health Act of 1970, Public Law 91-596, 84 Stat. 1590, signed December 29, 1970 from S. 2193 in the 91st Congress and effective April 28, 1971, codified at 29 U.S.C. 651 and following: the two-year startup shortcut that borrowed whole rulebooks from private standards organizations, the ordinary rulemaking process of section 6(b), the two Supreme Court decisions that together define what the agency must prove before regulating and what it need not, the emergency standard power whose record in the courts is poor enough that the agency uses it very rarely, and the enforcement chain that runs from inspection through citation to a contest decided by a body entirely separate from the agency that issued it.

The reader who finishes this article will be able to do four things that the standard summaries do not teach. First, trace each stage of the hazard-to-citation pipeline and name the actor, the legal test, and the timeline at each step. Second, state the holdings of Industrial Union Dep’t v. American Petroleum Inst., 448 U.S. 607 (1980), and American Textile Mfrs. Inst. v. Donovan, 452 U.S. 490 (1981), and explain why the two together make the agency prove the danger and then forbid it from trading that danger against cost. Third, explain why emergency temporary standards almost never survive review, by case outcomes rather than by characterization. Fourth, describe who actually decides a contested citation, and why that decider is not the agency that wrote the check it is being asked to cash.

This is the operational half of the cluster’s pillar. The pillar article maps the statute’s three-body institutional design, the general duty clause, and the coverage and penalty architecture; the full statutory foundation, the three institutions Congress created, and the catch-all duty that applies where no specific rule exists are laid out in the Occupational Safety and Health Act of 1970 guide. This article takes the machinery the pillar introduces and shows it running: how rules get written, how they get enforced, and where the courts have drawn the lines around both.

The namable claim of this article is short. Prove the risk, then ignore the cost. The two foundational decisions impose a significant-risk threshold and then bar cost-benefit balancing, which is an unusual and demanding combination. It explains both why new health standards are rare and why the ones that exist are stringent, and it explains the series thesis thread that runs through the whole cluster: the gap between the broad authority Congress granted in 1970 and the regulation actually delivered is real, and judicial doctrine sets the width of that gap.

Before the machinery, a note on method. Every claim in this article is tied to a section number, a case citation, or a dated event. Statutory sections are cited to the United States Code as codified, so section 6 of the act is 29 U.S.C. 655, section 8 is 29 U.S.C. 657, section 9 is 29 U.S.C. 658, section 10 is 29 U.S.C. 659, section 11 is 29 U.S.C. 661, section 12 is 29 U.S.C. 660, and section 17 is 29 U.S.C. 666. Case holdings are stated in the precise terms the opinions used, including which justices joined which parts, because the alignment of the Court in these cases is part of the story. The reference date for this article is March 15, 2015, and nothing in it describes any rule, dispute, or administration later than that date.

The Two-Year Shortcut That Wrote Most of the First Rulebook

The most startling fact about the body of enforceable workplace safety rules in the United States is how little of the original corpus went through rulemaking at all. When the Occupational Safety and Health Act took effect on April 28, 1971, the agency it created had no staff of rulemakers, no completed scientific record on the hazards it was supposed to control, and a congressional mandate to produce enforceable standards immediately. Congress solved this with section 6(a), codified at 29 U.S.C. 655(a), which gave the Secretary of Labor a two-year window, in the statute’s words ending two years after such date, to adopt as an occupational safety or health standard any national consensus standard and any established federal standard, without the notice, comment, and hearing procedures that section 6(b) otherwise requires. The provision was a startup device, and it was used wholesale.

On May 29, 1971, barely a month after the effective date, the Labor Department published a final rule at 36 FR 10466 that adopted national consensus standards and established federal standards en masse as enforceable occupational safety and health standards, effective in August 1971. The consensus standards came principally from the American National Standards Institute and the National Fire Protection Association, private bodies whose voluntary industry standards had been written through the 1960s. The established federal standards were rules already in force under earlier federal workplace safety laws, carried forward into the new regime. In a single rulemaking action, without public comment or hearings, the new agency acquired a full rulebook.

Why did Congress let a new agency adopt private standards without rulemaking?

Congress faced an agency with no scientific staff and a mandate to regulate immediately, so section 6(a) let the Secretary borrow credibility and speed by adopting existing consensus and federal standards wholesale during a two-year window. The tradeoff was durability: rules written as voluntary industry guidance in the 1960s became enforceable law.

The consequences of that choice are still visible in the structure of the rulebook. A substantial body of enforceable workplace rules traces to voluntary industry standards written in the 1960s, and some remain technologically dated because the shortcut froze the state of private standard-setting at the moment of adoption. A rule that began life as a voluntary ANSI recommendation about machine guarding or walking surfaces became, by the stroke of the May 1971 rule, a legal duty backed by civil penalties, without any agency ever having asked whether the 1960s consensus represented the best available protection or merely the protection the industry had been willing to write for itself. The shortcut also created a lasting asymmetry in the law: the safety standards, which govern physical hazards like falls, machines, and electricity, arrived largely complete on day one, while the health standards, which govern toxic substances and require exposure limits built on dose-response science the agency did not yet have, had to be built from scratch through the slow machinery of section 6(b). That asymmetry explains why the significant-risk and feasibility cases of 1980 and 1981 both concerned health standards rather than safety standards. The safety rulebook was borrowed; the health rulebook had to be earned, and the courts set the price of earning it.

The shortcut’s defenders and its critics read the same history differently, and both readings deserve a hearing because the tradeoff is genuine. The defense is that without section 6(a), the agency would have spent its first years with almost nothing enforceable, and workers in the most dangerous industries would have waited while the new department built a scientific apparatus from nothing. The criticism is that Congress laundered private, industry-influenced standards into public law without democratic procedure, and then left many of them in place for decades because the ordinary rulemaking process proved too slow to replace them. Both are true. The two-year window expired, the authority lapsed, and every standard written after it had to travel the full road of section 6(b). But the road the shortcut built is the one most employers still drive on: the great bulk of the safety standards an inspector cites in an ordinary factory or construction inspection descend from the May 1971 adoption, not from any rulemaking the agency conducted itself.

There is a recurring error worth correcting early, because it distorts every other part of the subject. It is the assumption that all OSHA standards were written through notice-and-comment rulemaking. They were not. The founding layer of the rulebook was adopted, not made, and understanding the difference matters for practitioners. A standard adopted under section 6(a) carries the same legal force as a standard written under section 6(b); an employer cited under either must answer the citation the same way. But the adopted standards rest on a thinner scientific and procedural record, which is one reason some of them have survived essentially unchanged while the science around them moved on. The agency’s later inability to update them quickly is not a separate failure; it is the second half of the same story, and the next section explains the machinery that made updating them so slow.

How a Rulemaking Begins

Before the notice, the comment period, and the hearing, there is the decision to begin, and the statute says less about that decision than practitioners might wish. Section 6(b) permits the Secretary to act on the Secretary’s own motion or on the petition of an interested person, which means a rulemaking can start from inside the agency or from outside it. A union that has watched members fall ill, an industry group that wants a uniform national rule to replace a patchwork of state requirements, a researcher whose new study documents a hazard, any of these may petition the agency to write a standard. The petition does not compel the agency to act; it compels the agency to consider, and the consideration happens inside a priority-setting process that is the least visible and most consequential stage of rulemaking.

The agency maintains a regulatory agenda, published twice yearly, listing the rulemakings it is working on and those it plans to begin. Placement on the agenda is the practical prerequisite to everything this article has described: no agenda slot, no risk assessment; no risk assessment, no proposal; no proposal, no standard. The agenda is where the agency rations its scarcest resource, which is not money but the scientific and legal staff time to build records that survive substantial-evidence review. A hazard that does not make the agenda does not get regulated, regardless of its severity, and the competition for agenda slots is the quietest form of priority-setting in the regulatory state.

What moves a hazard onto the agenda is a combination of science, politics, and litigation risk. New scientific evidence of a significant risk supplies the Benzene predicate. A petition with political support supplies the will. A court order or a congressional mandate supplies the compulsion, although mandates are rare in this field. What keeps a hazard off the agenda is the same calculus in reverse: the science is uncertain, the industry opposition is organized, the record would take a decade to build, and the enforcement tools already reach the hazard through existing standards or the general duty clause. The agenda is thus the place where the doctrinal constraints of this article cash out into institutional behavior. An agency that knows each new health standard requires a significant-risk finding, a feasibility showing, and a record that survives substantial-evidence review will put fewer health standards on its agenda, work them more slowly, and lean harder on enforcement of what already exists.

For the outside petitioner, the lesson is strategic. A petition that merely asserts a hazard exists will not move the agency; the agency already knows hazards exist. A petition that supplies the beginning of the record the agency would need, exposure data, epidemiological studies, evidence of feasible controls, and that frames the hazard in the language of the statute’s tests, significant risk and feasibility, gives the agency something it can use. The most effective petitions are those that do the agency’s early work for it, because the agency’s early work is the bottleneck. The rulemaking begins, in practice, not with the Federal Register notice but with the moment someone assembles enough evidence to make the agenda slot defensible, and that moment can come years before the public ever hears of the rule.

The Ordinary Road: How Section 6(b) Rulemaking Actually Works

Once the two-year window of section 6(a) closed, every new standard had to travel the ordinary road laid out in section 6(b), codified at 29 U.S.C. 655(b). The statute’s procedure is recognizably the notice-and-comment model used across federal regulation, with additions that reflect the technical character of workplace hazards. The agency publishes notice of a proposed rule in the Federal Register. Interested persons have 30 days after publication to submit written comments, a compressed comment period by the standards of modern rulemaking. Any interested person may request a hearing, and the statute requires the agency to hold one, with the opportunity for oral testimony and cross-examination on the issues the rule raises. At the end of the process, the agency must publish the final standard together with a statement of reasons for its action, required by section 655(b)(8) and section 655(e), which is the document a reviewing court will read to decide whether the rule survives.

That is the procedure on paper. In practice, the process routinely takes many years, and the reasons are structural rather than accidental. A health standard requires the agency to assemble a risk assessment, which means gathering exposure data across industries, evaluating epidemiological and toxicological studies, modeling dose-response relationships, and defending each choice against industry scientists who will testify at the hearing that the agency’s numbers are wrong. A safety standard requires engineering analysis of the controls the rule would mandate and evidence that those controls exist and work in the field. Both kinds of rules then face the two judicial tests this article examines in the next two sections: the significant-risk threshold and the feasibility requirement. Each test is an evidentiary burden the agency must carry on the rulemaking record, and each adds years, because the record must be built to survive a court that will examine it under the substantial-evidence standard of section 655(f), which allows a reviewing court to set aside a standard not supported by substantial evidence in the record considered as a whole.

The hearing stage deserves attention because it is where rulemaking time goes. Section 6(b) hearings are not the brief public listening sessions familiar from other agencies. They are extended, trial-like proceedings in which the agency’s proposed rule is defended witness by witness against organized opposition, with cross-examination and rebuttal. For a major health standard, the hearing record can run to thousands of pages of testimony and exhibits. The statement of reasons the agency writes afterward must answer the objections the hearing raised, because a court applying substantial-evidence review will look for the agency’s response to the contrary evidence. An agency that hurries this stage writes a record that a court vacates; an agency that does it carefully writes a record that takes years. There is no third option, and that is the first reason the pace of new health standards slowed so dramatically after the early 1980s.

How long does it take to write an OSHA standard?

There is no statutory deadline, and the record shows the honest answer is measured in years rather than months. Building the scientific record, holding trial-like hearings with cross-examination, answering objections in the statement of reasons, and surviving substantial-evidence review in court together make major standards multi-year undertakings by structural necessity.

The second reason is resources, and it is worth stating plainly because it is often left out of doctrinal accounts. Rulemaking of this kind is expensive in staff time and scientific capacity. The agency must maintain the toxicologists, epidemiologists, engineers, and economists to build and defend the record, and it must do so while also running the inspection and enforcement program that consumes most of its field personnel. A rulemaking docket competes for the same leadership attention as the enforcement program, and enforcement, being visible and immediate, usually wins the contest for priority. The result is a pipeline in which only a few major standards can move at once, each moving slowly.

The third reason is the one the courts supplied, and it is the subject of the next two sections. In 1980 and 1981, the Supreme Court handed down the two decisions that define the evidentiary burdens of health standard rulemaking. The first added a threshold the agency must cross before it may regulate at all. The second removed the defense the agency might otherwise have used to limit how far it must go. Together they made the rulemaking record harder to build and the resulting standard harder to moderate, which is exactly the combination that produces few rules, each of them demanding.

A note on a related mechanism before leaving section 6. The statute also provides for variances in section 6(d), codified at 29 U.S.C. 655(d), under which an employer that cannot comply with a standard by its effective date, or that uses an alternative method providing equal or better protection, may apply for permission to deviate. The variance process is the safety valve of the standards system: it acknowledges that a uniform national rule will fit some workplaces badly, and it gives those employers a lawful path other than violation. In practice it is used sparingly, because the application requires the employer to demonstrate equivalent protection, which is itself an evidentiary undertaking. The variance exists, it is real law, and it is one of the least invoked provisions in the statute, which tells its own story about how employers actually experience the standards.

The research function that feeds rulemaking sits deliberately outside the rulemaking agency. Congress placed the National Institute for Occupational Safety and Health in the health agency rather than the Labor Department, so that the scientific recommendations informing standards would not be produced by the same body that must defend them in litigation. The institute develops criteria documents recommending exposure limits and control measures, and the rulemaking agency draws on that work when it builds the record for a health standard. The separation is part of the three-body design the pillar article maps: research, enforcement, and adjudication kept apart so that no single institution both finds the hazard and conclusively punishes the employer. For the rulemaking reader, the practical point is that the scientific starting point of a standard often comes from outside the agency that writes it, which strengthens the credibility of the science and complicates the politics of the rule, since the rulemakers cannot disown the researchers and the researchers cannot be ordered to soften their findings.

The First Constraining Decision: Prove the Danger

In 1980 the Supreme Court decided Industrial Union Dep’t v. American Petroleum Inst., 448 U.S. 607 (1980), known in the field as the Benzene case, and added a threshold burden to health standard rulemaking that had not been visible in the statute’s text. The case concerned the agency’s standard for benzene, a toxic substance used across the chemical and refining industries. The agency had set its exposure limit at the lowest level it considered feasible. A plurality of the Court, in an opinion by Justice Stevens joined by Chief Justice Burger and Justice Stewart, with Justice Powell joining in the relevant parts, held that before the agency may regulate a toxic substance, it must first find that a significant risk exists. The judgment was affirmed 5 to 4. Justice Marshall dissented, joined by Justices Brennan, White, and Blackmun. The question whether the statute requires cost-benefit balancing was expressly reserved, left for another day and another case.

The holding needs careful parsing because it is frequently overstated in both directions. The plurality did not say the agency must prove harm with scientific certainty, which would have made health regulation nearly impossible, since the science of low-dose exposure rarely supplies certainty. It said the agency must make a threshold finding, supported by the record, that the workplace exposure presents a significant risk of material health impairment, before the agency may use section 6(b) to regulate the substance. The word significant does the work. It means the risk must be more than theoretical, more than a mere possibility, and substantial enough to justify the machinery of regulation. It does not mean the agency must quantify the risk to a precise number of expected illnesses, but it does mean the agency cannot regulate on speculation, and it must show its work.

The practical effect was to add a gate at the entrance to rulemaking. Before Benzene, the agency could reasonably read section 6(b) as authorizing it to set the most protective feasible limit for any toxic substance found in the workplace, with the feasibility analysis doing the real constraining work. After Benzene, the agency had to prove the danger first, as a separate evidentiary showing, and only then move to the question of how protective the standard must be. For the rulemaking record, this meant building a risk assessment capable of withstanding judicial scrutiny as a distinct element, with exposure data, epidemiological evidence, and a reasoned explanation of why the risk qualifies as significant. For the timeline, it meant more years, because risk assessment of this kind is the slowest part of the science.

What must OSHA prove before regulating a toxic substance?

The agency must make a threshold finding, supported by the rulemaking record, that workplace exposure to the substance presents a significant risk of material health impairment. It need not prove harm with scientific certainty, but it cannot regulate on speculation; the risk must be real, substantial, and demonstrated with evidence.

The alignment of the justices matters for understanding what the decision did and did not settle. A plurality opinion, even one whose judgment commands five votes, carries less precedential weight than a majority opinion, and the Benzene plurality’s reasoning about how significant a risk must be has been debated ever since. What stuck, and what every subsequent health rulemaking has had to satisfy, is the threshold itself: no significant-risk finding, no regulation. The dissent’s warning, in essence, was that the threshold would become a tool for delay, requiring the agency to perfect its science before acting while workers continued to be exposed. The plurality’s answer, in essence, was that Congress did not authorize the agency to regulate risks it had not shown to be significant, and that the agency’s expertise does not excuse it from proving the predicate of its own power. Both positions have echoed through every health rulemaking since, and the tension between them is one of the standing arguments of the field.

There is a subtle point about the decision’s scope that practitioners sometimes miss. The significant-risk requirement applies to health standards governing toxic substances, where the agency regulates on the basis of risk assessment. It does not apply in the same way to safety standards governing physical hazards, where the danger is typically observable and the question is what control the rule requires. This is another facet of the asymmetry the startup shortcut created: the safety rulebook, largely borrowed, mostly escapes the threshold, while the health rulebook, built from scratch, must cross it every time. The result is that the most scientifically demanding part of the workplace safety program faces the highest evidentiary bar, which is one reason the health standards are both the most protective and the rarest.

The decision also left the cost question open, and the reservation was explicit. The plurality did not decide whether the statute permits or requires the agency to balance costs against benefits, because the case could be resolved on the threshold ground. That left the agency, the industries, and the lower courts in an unstable position: the agency had to prove significant risk, but nobody knew whether, having proved it, the agency could then be forced to show that the benefits of regulation justified the costs. The answer came one year later, and it came as a surprise to those who had expected the Court to complete the Benzene framework with a cost-benefit requirement.

The Second Constraining Decision: Ignore the Cost

In 1981 the Supreme Court decided American Textile Mfrs. Inst. v. Donovan, 452 U.S. 490 (1981), the Cotton Dust case, and answered the question Benzene had reserved. The case concerned the agency’s standard for cotton dust in textile mills, a respiratory hazard of long standing in that industry. The industry’s argument was, in substance, that the agency should have balanced the standard’s costs against its benefits and chosen a less demanding limit. The Court rejected that argument. By 5 to 3, in an opinion by Justice Brennan joined by Justices White, Marshall, Blackmun, and Stevens, with Justice Powell taking no part, the Court held that the statute does not require cost-benefit balancing. Justice Stewart dissented. Justice Rehnquist dissented, joined by Chief Justice Burger. The statute requires feasibility, and feasibility alone.

The holding turns on the language of section 6(b)(5), codified at 29 U.S.C. 655(b)(5), which directs the agency, in setting standards for toxic substances, to adopt the standard that most adequately assures, to the extent feasible, that no employee will suffer material impairment of health. The Court read that phrase as Congress’s answer to the cost question. Feasible means capable of being done, technologically and economically. The agency must therefore adopt the most protective standard that is technologically and economically achievable. It may not stop short of that point on the ground that the last increment of protection costs more than it is worth in quantified benefits, and it may not be forced to go beyond that point into the infeasible on the ground that the benefits would justify it. Cost enters the analysis only through the feasibility door: a standard that would bankrupt the industry or require technology that does not exist is not feasible, and the agency may not impose it. But within the feasible range, the agency must maximize protection, not optimize the ratio of benefits to costs.

This is a more demanding protection standard than most regulatory statutes impose, and the comparison is instructive. Many regulatory schemes instruct the agency to balance costs and benefits, or to regulate where the benefits justify the costs, which gives the agency discretion to stop when the next dollar of protection buys less than a dollar of benefit. The workplace safety statute does the opposite. Once the agency has crossed the Benzene threshold by showing significant risk, the Cotton Dust rule forbids it from trading the remaining risk against cost. The agency must push protection as far as feasibility allows, even if a cost-benefit analysis would have stopped earlier. The combination is unusual in American regulatory law, and it is the reason the namable claim of this article holds: prove the risk, then ignore the cost.

Must OSHA balance a standard’s costs against its benefits?

No. The Supreme Court held in 1981 that the statute requires feasibility, not cost-benefit balancing. The agency must adopt the most protective standard that is technologically and economically achievable, and it may not stop short of that point because the remaining protection costs more than quantified benefits would justify.

The dissents are worth reading because they state the counter-model plainly. The dissenting justices argued, in essence, that Congress could not have intended the agency to impose enormous costs for marginal gains in protection, and that a reasonable reading of the statute would allow the agency to consider whether the game is worth the candle. The majority’s answer was textual: Congress wrote feasibility, not cost-benefit balancing, into the statute, and the Court would not rewrite the bargain. The debate maps directly onto the standing political argument about the agency, with business constituencies invoking the dissent’s logic and labor constituencies invoking the majority’s, but the law is the majority’s, and it has governed every health rulemaking since 1981.

The feasibility standard has its own internal structure that practitioners must understand. Technological feasibility asks whether the controls the standard requires exist and can be implemented in the affected industries, or can reasonably be developed. Economic feasibility asks whether the affected industries can absorb the costs without massive dislocation, a test the courts have generally read as forgiving of significant cost but not of industry-wide destruction. The agency bears the burden of showing feasibility on the rulemaking record, and the substantial-evidence standard of section 655(f) applies. In practice, this means the rulemaking record must contain industry-by-industry evidence on costs and capabilities, which is the second great consumer of rulemaking time after the risk assessment. The agency must prove the danger with science and prove the achievability with economics, and it must do both on a record that a hostile court will examine for substantial evidence.

There is a recurring error that this decision corrects, and it is one of the most common misunderstandings in the field. Many readers assume, from general familiarity with regulatory policy, that cost-benefit analysis is required for all major federal regulation. For this statute, it is not. The assumption usually comes from knowledge of executive orders requiring cost-benefit review of agency rulemaking generally, and it is true that such review exists as a matter of executive policy. But the statute itself, as construed by the Court, forbids the agency from using cost-benefit balancing as the decisional standard. The distinction between the statutory test and executive-branch review procedures is one of the points that makes this body of law standard examinable material in administrative and employment law coursework, and students working through the two holdings often test their understanding against the VaultBook legislation study notebook alongside the statute’s text.

Read together, the two decisions produce the regime the agency has operated under for more than three decades. Benzene makes the agency prove the danger. Cotton Dust forbids the agency from trading the proven danger against cost. The first decision raised the bar for starting; the second removed the brake on finishing. The predictable result is exactly what the record shows: new health standards are rare, because the threshold is hard to cross and the record is hard to build, and the standards that do issue are stringent, because feasibility, not cost-benefit optimization, sets their stringency. A reader who understands those two sentences understands why the health side of the rulebook looks the way it does, and why the agency’s practical power shifted, after 1981, toward the enforcement tools that do not require new rulemaking at all.

Reading “Most Adequately Assures, to the Extent Feasible”

The most consequential sentence in the standard-setting title is section 6(b)(5), and it repays the close reading the series’ statute-reading method demands. In the statute’s words, the Secretary, in promulgating standards for toxic materials or harmful physical agents, shall set the standard which most adequately assures, to the extent feasible, on the basis of the best available evidence, that no employee will suffer material impairment of health or functional capacity even if such employee has regular exposure to the hazard for the period of his working life. Every phrase in that sentence carries weight, and the Cotton Dust decision turned on the Court’s reading of the whole.

“Most adequately assures” is a superlative, not a balancing test. It does not ask the agency to assure adequately, or reasonably, or cost-effectively. It asks for the most adequate assurance available, which pushes the agency toward maximum protection as the starting point. “To the extent feasible” is the only modifier, and it operates as a ceiling rather than a scale: feasibility limits how far the agency must go, but nothing in the phrase authorizes the agency to stop short of the feasible maximum for cost-benefit reasons. “On the basis of the best available evidence” tells the agency what to build the standard on, the best evidence available, not perfect evidence, which matters because the science of toxic exposure is almost never complete. “No employee will suffer material impairment” states the protective ambition in absolute terms, and “even if such employee has regular exposure for the period of his working life” extends the protection across a full working lifetime, which is why the standards must account for chronic, low-dose exposure rather than only acute effects.

The Court in Cotton Dust read this sentence as Congress’s deliberate choice of a feasibility standard over a cost-benefit standard, and the textual analysis supports the reading. If Congress had wanted cost-benefit balancing, it knew how to say so; other statutes of the era say it explicitly. Instead, Congress wrote a superlative protection mandate limited only by feasibility, and the Court declined to rewrite the sentence. The dissent’s position, that the sentence must implicitly allow consideration of the relationship between costs and benefits, foundered on the text’s plain direction: the agency is told what to maximize, protection, and what constrains the maximization, feasibility, and cost-benefit balancing is neither the maximand nor the constraint.

For the rulemaking practitioner, the sentence structures the entire evidentiary enterprise. “Best available evidence” governs the risk assessment: the agency must use the best science it has, document its choices among conflicting studies, and explain why the chosen evidence is the best available. “To the extent feasible” governs the economic and technological analysis: the agency must show, industry by industry, that the standard can be achieved with available or reasonably developable technology and without destroying the industry’s economic viability. “Most adequately assures” governs the final choice of the exposure limit: among the feasible alternatives, the agency must choose the most protective. The sentence is thus not merely a standard of review for courts. It is the rulemaking’s table of contents, and a well-built rulemaking record follows its order: the evidence, the feasibility, the protective choice.

How Courts Read the Rulemaking Record

The substantial-evidence standard of section 655(f) is the lens through which every standard is judged, and it deserves a section of its own because it shapes rulemaking behavior more directly than any other single provision. When the agency issues a standard under section 6(b), any person adversely affected may challenge it in a federal court of appeals, and the court must set aside the standard if it is not supported by substantial evidence in the record considered as a whole. That sentence contains three terms of art, and each one does work.

Substantial evidence means more than a scintilla but less than a preponderance: enough evidence that a reasonable mind could accept as adequate to support the conclusion. The record considered as a whole means the court looks at everything, including the evidence that cuts against the agency’s conclusion, not just the evidence the agency chose to emphasize. And the challenge goes directly to the court of appeals, not through the commission, because a standard is a legislative act of the agency rather than an enforcement action against a particular employer. The practical consequence is that the rulemaking record is built from the beginning with an appellate audience in mind. Every study cited, every witness called, every objection answered in the statement of reasons is there because a panel of judges will eventually read the record looking for the contrary evidence and asking whether the agency dealt with it.

The standard is more demanding than the arbitrary-and-capricious review that governs most agency rulemaking, and the difference matters. Under arbitrary-and-capricious review, a court asks whether the agency considered the relevant factors and drew a rational connection between the facts and the choice. Under substantial-evidence review, the court examines the evidentiary foundation itself, weighing whether the record as a whole supports what the agency did. For the agency’s lawyers, this means the rulemaking cannot rest on expert judgment alone; it must rest on evidence in the record that a skeptical judge can see and evaluate. For the agency’s scientists, it means every risk assessment must be documented to a standard that survives adversarial testing. For the timeline, it means the record-building stage cannot be hurried, because a thin record is a vacated standard.

The interaction between substantial-evidence review and the two Supreme Court decisions creates the evidentiary architecture of modern health rulemaking. Benzene requires the record to contain the significant-risk finding, with the exposure data and epidemiological evidence to support it. Cotton Dust requires the record to contain the feasibility showing, with the technological and economic evidence industry by industry. Section 655(f) requires both showings to survive a court’s examination of the whole record, including the contrary evidence the hearing produced. An agency planning a major health standard therefore faces a record-building task with three evidentiary pillars, each of which must stand on its own, and each of which the industry’s experts will attack at the hearing and again before the court. The multi-year timeline is not a sign of agency torpor. It is the time required to build a record that can carry three pillars past a court that is instructed to look at all of the evidence.

There is a strategic consequence that experienced practitioners understand well. Because the standard of review is so demanding, the agency has a strong incentive to build rulemaking records that over-prove their case, adding studies, analyses, and responses to objections beyond what the statute strictly requires. This over-proving lengthens the rulemaking further, but it is rational insurance against vacatur. A vacated standard costs the agency years of work and emboldens the next challenge; a slow standard that survives costs only time. The agency’s revealed preference, across decades, has been for the slow survivor over the fast casualty. Critics of the agency read this as foot-dragging. Defenders read it as prudence under a demanding standard of review. Both readings describe the same behavior, which is the behavior the statute’s review provision rewards.

Emergency Standards: The Power the Agency Almost Never Uses

Section 6(c), codified at 29 U.S.C. 655(c), gives the agency its most dramatic power and its least successful one. When the Secretary finds that employees are exposed to a grave danger from a hazard, and that an emergency standard is necessary to protect them, the agency may issue a temporary standard that takes effect immediately, without the notice, comment, and hearing procedures of section 6(b). The emergency standard serves as the proposed rule for the ordinary proceeding that must follow, and it expires after six months unless replaced by a permanent standard issued through section 6(b). The design is plainly meant for true emergencies: a newly discovered carcinogen, a sudden pattern of fatalities, a hazard for which waiting years would cost lives.

The record of the power in the courts is poor, and the numbers tell the story without characterization. According to the Congressional Research Service’s compilation in CRS R46288, Table A-1, the agency issued exactly 9 emergency temporary standards between 1971 and 1983, and none at all from 1984 through March 2015. Of the 9, the outcomes were: asbestos in 1971, unchallenged; organophosphorous pesticides in 1973, vacated by the courts; 14 carcinogens in 1973, with 12 upheld and 2 vacated; vinyl chloride in 1974, unchallenged; diving operations in 1976, stayed by the courts; benzene in 1977, stayed; DBCP in 1977, unchallenged; acrylonitrile in 1978, where the court denied a stay; and asbestos in 1983, stayed. Most of the contested emergency standards were stayed or vacated. The agency’s experience taught a lesson the numbers make unavoidable, and the agency learned it: after 1983, the power sat unused for more than three decades.

The reasons for the poor record are structural, and they follow from the nature of emergency action. A court reviewing an emergency standard asks whether the agency has shown a grave danger with evidence sufficient to justify bypassing the ordinary procedures. But the ordinary procedures are precisely the machinery that produces evidence: notice, comment, hearings, and the statement of reasons. An agency acting on an emergency timeline has, by definition, a thinner record than an agency that took years, and the grave-danger finding must be made on that thinner record. The courts have applied exacting review to that finding, and the combination of a thin record and exacting review is why most emergency standards have been stayed or vacated. The agency can act fast, or it can act on a record that survives review, and the statute’s six-month clock does not give it time to do both.

This dynamic belongs to a longer judicial story about the limits of agency claims to emergency authority. The courts have grown increasingly willing to check agency action taken without clear congressional authorization, a line of decisions that has reached emergency workplace standards among other fields. The series traces that line of authority through the Massachusetts v. EPA explainer, which maps how the Court polices the boundary between agency discretion and congressional delegation. The emergency standard record is the early chapter of that story in the workplace safety field: the courts did not strike the power from the statute, but they set the evidentiary price of using it so high that the agency stopped paying it.

For the practitioner, the lesson is concrete. An emergency temporary standard is not a shortcut around the Benzene and Cotton Dust requirements. The grave-danger finding is, if anything, a more demanding evidentiary showing than the significant-risk finding, because it must justify immediate effectiveness without the procedural protections of section 6(b), and it must be made quickly. The six-month expiration means the agency must simultaneously defend the emergency standard in court and complete the ordinary rulemaking to replace it, fighting on two fronts with one staff. The rational agency response to those incentives is the one the record shows: reserve the power for the rare case where the danger is so grave and so well documented that the finding can survive, and otherwise do the slow work of section 6(b). The emergency standard is the power the statute advertises most loudly and the agency uses most rarely, and the gap between the advertisement and the use is judicial doctrine’s doing.

There is a final irony worth noting. The emergency standard was designed for the situation in which waiting would cost lives, yet the judicial record has made waiting the safer legal course. An agency leader who believes workers face a grave danger must weigh the lives at stake against the near certainty that a court will stay the standard, leaving the agency with the worst of both worlds: no protection in place and a litigation loss that emboldens the next challenge. That calculus, repeated across administrations of both parties, is the institutional explanation for the three-decade silence. The power remains on the books, the six-month clock remains in the statute, and the record remains the warning.

Nine Emergencies: What Each One Taught

The nine emergency temporary standards, taken one by one, are the case law of the emergency power, and they read as a education in what the courts demand. The Congressional Research Service’s Table A-1 in CRS R46288 records all nine, issued between 1971 and 1983, with no emergency standard issued from 1984 through March 2015. Each outcome below is stated as the record states it, without characterization, because the pattern needs no adornment.

Asbestos, 1971, was unchallenged. The first emergency standard in the statute’s history drew no lawsuit, which tells us something about the early atmosphere: the hazard was well documented, the industry’s capacity to contest was not yet organized, and the new agency’s assertion of emergency authority went untested. Organophosphorous pesticides, 1973, was vacated. The court’s vacatur meant the standard was set aside as unlawful, the strongest judicial rebuke available, and it put the agency on notice that the grave-danger finding would be examined rather than assumed. The 14 carcinogens standard of 1973 produced a split outcome: 12 of the 14 substance provisions were upheld and 2 were vacated. The partial survival is instructive. It shows the courts were willing to sustain emergency action where the record supported it substance by substance, and willing to carve out the provisions where it did not. For the agency’s lawyers, the lesson was that an emergency standard covering multiple substances would be reviewed provision by provision, and the weakest link would not sink the whole rule but would be severed from it.

Vinyl chloride, 1974, was unchallenged, the second emergency standard to go into effect without litigation. Diving operations, 1976, was stayed: the court blocked the standard from taking effect while the challenge proceeded, which for a six-month emergency standard is often the practical equivalent of vacatur, since the clock runs while the litigation does. Benzene, 1977, was stayed as well, and the benzene story continued: the emergency standard was blocked, the agency returned to the ordinary process of section 6(b), and the permanent benzene standard that emerged became the subject of the Supreme Court’s 1980 decision. The sequence from stayed emergency standard to litigated permanent standard to significant-risk threshold is the single most consequential chain of events in the history of health rulemaking, and it began with a court refusing to let the emergency version take effect.

DBCP, 1977, was unchallenged. Acrylonitrile, 1978, produced the agency’s rare procedural victory: the court denied the request for a stay, allowing the emergency standard to remain in effect during the challenge. The denial of a stay is worth pausing over, because it shows what a successful emergency record looks like. The agency had documented the danger well enough, and the challengers’ showing of likely success weak enough, that the court let the standard operate. It is the exception that illuminates the rule: most emergency records did not clear that bar. Asbestos, 1983, the ninth and, through the March 2015 horizon, last emergency temporary standard, was stayed. After that stay, the agency did not try again for more than three decades.

Several patterns emerge from the nine that no single case shows. First, the unchallenged standards, asbestos in 1971, vinyl chloride in 1974, DBCP in 1977, cluster in the early and middle years, before the regulated community had fully learned to litigate emergency action. Second, the stayed standards cluster later, as judicial skepticism hardened and challengers learned that a stay motion against a six-month standard is often dispositive. Third, the vacaturs, organophosphorous pesticides in full and 2 of the 14 carcinogens in part, show courts willing to reach the merits and set aside what the record did not support. Fourth, the single denied stay, acrylonitrile, shows the bar was clearable but rarely cleared. Taken together, the nine form a record in which the agency won outright only when no one sued or when its record was unusually strong, and lost or stalled whenever a challenger forced a court to examine the grave-danger finding closely.

The institutional memory of these nine shaped everything after 1983. Agency leaders considering emergency action did not need a legal memorandum to understand the odds; they needed only the table. An emergency standard would be challenged, the challenger would seek a stay, the court would examine the grave-danger finding on a thin emergency record, and the likely outcome was a stay or vacatur that left the agency worse off than if it had begun the ordinary rulemaking immediately. The rational response was the one the record shows: keep the power on the books, and never use it. The three-decade silence is not a mystery. It is the predictable behavior of an institution that learned, case by case, what the courts would do.

Inspections: The Warrant Rule and How the Agency Gets in the Door

Standards without inspections are advice. Section 8 of the act, codified at 29 U.S.C. 657, gives the agency the inspection power that turns the rulebook into an enforcement program. Compliance officers may enter workplaces, inspect conditions, question employees, and review records, and the inspections are, by design, unannounced: advance notice of an inspection is itself a violation, because an employer warned of an inspector’s coming could stage compliance for the day. The inspection is the point of first contact between the regulatory state and the regulated workplace, and the law governing that contact was settled by the Supreme Court in 1978.

In Marshall v. Barlow’s, Inc., 436 U.S. 307 (1978), the Court held that the agency may not inspect a workplace over the employer’s objection without a warrant or the functional equivalent. The vote was 5 to 3, in an opinion by Justice White; Justice Stevens dissented, joined by Justices Blackmun and Rehnquist; Justice Brennan took no part. The warrant the decision requires is not the criminal warrant of television drama. It is an administrative warrant issued on a showing of administrative probable cause, which the Court defined to include either specific evidence of a violation or a showing that the workplace was selected under reasonable administrative standards for inspection. The decision constitutionalized the employer’s right to insist on that showing before the inspector crosses the threshold.

The practical operation of the rule is less dramatic than the holding sounds. Most inspections proceed by consent: the compliance officer arrives, presents credentials, and the employer allows entry, often after calling counsel. The warrant requirement matters at the margin, where an employer refuses entry, and it matters structurally, because it forces the agency to articulate, in advance and on paper, why this workplace was selected. The administrative probable cause standard is deliberately easier to satisfy than criminal probable cause, and courts have upheld warrants based on neutral inspection programs, employee complaints, and industry targeting plans. But the requirement is real, and it is one of the judicial doctrines that set the width of the gap between statutory authority and delivered regulation: Congress wrote an inspection power that reads as plenary, and the Court read the Fourth Amendment to require a warrant first.

The inspection itself follows a choreography the statute and the agency’s procedures have refined over decades. The opening conference tells the employer why the inspector is there. The walkaround takes the compliance officer through the workplace, and section 657(e) gives employees the right to have a representative accompany the inspector, the walkaround right that makes the union representative or the designated employee delegate a participant in enforcement rather than a spectator. The closing conference tells the employer what the inspector found. Between the walkaround and the closing conference, the inspector documents conditions, interviews employees privately, and reviews injury logs and other records. The documentation matters because everything that follows, the citation, the contest, the commission, the court, will be decided on the record the inspection created.

What the agency inspects, and how often, is a matter of institutional priorities rather than universal coverage. The agency has never had the personnel to inspect every workplace, or anything close to it, so inspections are targeted: programmed inspections aimed at high-hazard industries and workplaces with poor injury records, and unprogrammed inspections responding to employee complaints, referrals, and fatalities or catastrophes. Enforcement intensity varies with administrations, and that variation is a standing political argument, with business constituencies reading inspection numbers as overreach and labor constituencies reading the same numbers as neglect. The institutional fact underneath the argument is the one the statute created: a finite inspection force, a vast regulated universe, and a targeting system that rations attention by hazard and history. Any honest account of enforcement must describe the rationing, because the rationing is the policy.

The complaint-driven side of the system deserves emphasis because it is the most accessible enforcement tool for the workers the statute protects. An employee who believes a workplace violates a standard or presents a recognized hazard may file a complaint with the agency, and the statute protects the complaining employee against retaliation. Section 11(c), codified at 29 U.S.C. 660(c), provides that an employee who believes they have been discharged or discriminated against for exercising rights under the act may file a complaint with the Secretary within 30 days, and the Secretary may then bring an action in federal district court for appropriate relief. The 30-day deadline is short, and it is one of the provisions practitioners most often wish their clients had known about sooner. The whistleblower mechanism is the individual counterpart to the inspection program: where inspections ration the agency’s attention by program, complaints direct it by human report, and the anti-retaliation provision tries to make the reporting safe.

A note on records is necessary because inspections live or die on paper. Employers in covered industries must maintain logs of work-related injuries and illnesses, and the inspector’s review of those logs is often where an inspection’s real findings begin. The logs are also the data source for the targeting programs, which means the recordkeeping requirement serves enforcement twice: once as a substantive duty whose violation can itself be cited, and once as the intelligence that directs the inspectors. An employer who keeps sloppy logs invites the attention the logs are meant to satisfy, which is the kind of practical paradox that fills the working life of safety professionals.

Administrative Probable Cause in Practice

The warrant rule of Marshall v. Barlow’s is stated in a sentence; its practice fills manuals. Administrative probable cause, the showing the agency must make to obtain an inspection warrant over an employer’s objection, can be satisfied in two ways, and the two ways reveal how the enforcement system actually targets workplaces. The first is specific evidence of a violation: an employee complaint describing a hazard, a referral from another agency, a report of a fatality or catastrophe. The second is a showing that the workplace was selected under reasonable administrative standards: a neutral inspection program aimed at high-hazard industries, a follow-up on a prior violator, a program focused on a particular hazard across an industry. What does not satisfy the standard is the inspector’s whim, a hunch without grounding, or selection for an impermissible reason.

The ex parte warrant application is the procedural form this takes. The agency applies to a federal magistrate judge without the employer present, presenting an affidavit that sets out either the specific evidence or the administrative plan under which the workplace was selected. The magistrate’s review is not a rubber stamp, but it is deferential to reasonable administrative standards, and most applications are granted. The employer’s remedy, if the warrant was improperly issued, comes later: in the contest proceeding before the commission, the employer may challenge the validity of the inspection, and evidence obtained in violation of the warrant requirement may be suppressed. The suppression remedy is what gives the warrant rule its practical bite. An agency that cuts corners on the warrant risks losing not just the inspection but the citation built on it.

For the compliance officer in the field, the warrant process is the exception rather than the rule, and the consent inspection is the norm. The officer arrives unannounced, presents credentials, and requests entry. Most employers consent, often after a brief consultation with counsel, because refusing entry triggers the warrant application, which the agency will usually win, and the refusal itself signals to the agency that the workplace has something worth a second look. The strategic calculus for employers is therefore to consent and manage the inspection rather than to refuse and litigate the warrant, reserving warrant challenges for the rare case where the agency can point to no hazard, complaint, or inspection plan behind the selection. For the agency, the calculus runs the other way: the warrant application costs time and attorney effort, so the field office prefers the consenting employer and reserves compelled inspections for workplaces where consent is refused and the evidence of a hazard justifies the effort.

The constitutional dimension of the rule extends beyond the entry itself. Once inside, the scope of the inspection must be reasonable in relation to its justification: an inspection predicated on a complaint about a specific machine is not a license to rummage through every corner of the plant, although an inspector who observes a violation in plain view during a lawful inspection may cite it. The walkaround right of section 657(e), which lets an employee representative accompany the inspector, operates within this framework as well: the representative’s presence is a statutory right, not a matter of the employer’s grace, and interference with it can itself become an issue in the enforcement action. The inspection is thus a legally structured encounter from the moment of entry to the closing conference, with rights on both sides and a record being built for the adjudication to come.

From Inspection to Citation to Contest: Who Decides

The inspection produces findings. Section 9, codified at 29 U.S.C. 658, converts findings into citations. When the inspector believes an employer has violated a standard or the general duty clause, the agency issues a citation describing the violation with reasonable promptness, as the statute phrases it, and notifies the employer of the proposed penalty. The citation states what was violated, where and when, and what the employer must do to abate the hazard, with a deadline for abatement. The proposed penalty follows the classification of the violation, which the next section details. The citation is the agency’s formal accusation, and everything after it is adjudication.

Section 10, codified at 29 U.S.C. 659, gives the employer 15 working days from receipt of the citation to contest it. The deadline is jurisdictional in effect: an employer that misses it loses the right to challenge the citation, and the citation becomes a final order by operation of law. The contest right belongs to the employer, and employees and their representatives may also contest the reasonableness of the abatement period. The 15-day clock is the single most important deadline in the enforcement system, and employment lawyers treat it accordingly: the first advice to a client holding a new citation is to calendar the contest date before reading the citation’s substance.

Who decides whether an OSHA citation stands?

Not the agency that issued it. A contested citation goes to the Occupational Safety and Health Review Commission, an independent body of three presidential appointees entirely separate from the agency and the Labor Department, and from there to a federal court of appeals. The prosecutor does not sit as the judge.

The commission is the institution almost no one outside the field knows exists, and it is the reason the enforcement system has the shape it does. Section 11 of the act, codified at 29 U.S.C. 661, creates the Occupational Safety and Health Review Commission as an independent adjudicatory body composed of three members appointed by the President. The statute and the commission’s own description make the independence explicit: the commission is entirely separate from the agency that writes the standards and conducts the inspections, and entirely separate from the Department of Labor that houses that agency. When an employer contests a citation, the case goes first to an administrative law judge of the commission, who hears evidence and issues a decision, and then, on review, to the commission itself. The agency appears as a party, bearing the burden of proving the violation, and the employer defends. The structure is deliberately prosecutorial on one side and judicial on the other, with the two functions assigned to different institutions.

This is the point the recurring error gets wrong. Many readers assume the agency decides its own contests, as agencies often do in systems where the same body investigates, charges, and adjudicates. Here Congress deliberately split the functions. The agency that issues the citation does not decide whether it stands; the independent commission does, and the commission owes the agency’s interpretation no special deference as a party. The design is the adjudicatory half of the three-body architecture the pillar article maps, and it explains why enforcement outcomes are so often litigated rather than imposed. An employer facing a citation faces a prosecutor with a case to prove, not a regulator with a decision to announce, and the difference shapes every settlement negotiation in the system.

From the commission, section 12, codified at 29 U.S.C. 660, provides judicial review in the federal courts of appeals. Any person adversely affected by a commission order, which in practice means the employer or the Secretary, may seek review within 60 days. The court of appeals reviews the commission’s factual findings under the substantial-evidence standard and its legal conclusions de novo, which means the court decides for itself what the statute and the standards mean. For standards themselves, as distinct from citations, section 655(f) provides pre-enforcement review in the courts of appeals under the same substantial-evidence standard that governs the rulemaking record. The appellate stage is where the doctrinal lines drawn in Benzene, Cotton Dust, and Barlow’s are applied to new facts, and it is where the regulated community’s challenges to both rules and citations are finally resolved.

The adjudication structure differs from the single-agency model common in other regulatory fields, and the contrast illuminates both. In the environmental field, for example, the same agency typically writes the rule, enforces it, and adjudicates violations administratively, with courts reviewing the agency’s own decisions. The series’ treatment of EPA rulemaking under the Clean Air Act shows that contrasting regime in operation, where rulemaking volume is high and the agency’s interpretive authority is concentrated. The workplace safety statute chose separation where the environmental statutes chose concentration, and the choice has consequences: separated adjudication is slower and more litigation-intensive, but it gives the regulated community a neutral forum and denies the enforcement agency the home-field advantage. Neither design is neutral in its effects, and the choice between them is one of the durable institutional decisions of 1970 that still governs how the law performs.

The timeline of a contested case is measured in months and often years. The 15 working days to contest, the months for the administrative law judge proceeding, the further months for commission review, and the year or more for court of appeals review add up to an enforcement action whose final resolution may come long after the hazard was abated. The delay has a perverse effect worth naming: because abatement is typically required during the contest, the employer’s incentive is often to fix the hazard promptly and fight about the penalty and the characterization later, which means the system’s deterrent signal arrives long after its protective effect. Practitioners understand this rhythm; outsiders who expect enforcement to work like a traffic ticket, violation then swift fine, misunderstand the institution.

How the Commission Tries a Contested Case

The contest proceeding before the Occupational Safety and Health Review Commission is a trial, and it should be understood as one. When an employer files a timely notice of contest, the case is docketed before an administrative law judge of the commission. Discovery follows, in a form adapted from federal civil practice: document requests, interrogatories, depositions. The hearing itself features opening statements, witness testimony under oath, cross-examination, and documentary exhibits, all on a record. The Secretary of Labor appears as a party, represented by the Labor Department’s attorneys, and bears the burden of proving the violation. The employer defends, represented by counsel of its choice. The judge issues a written decision with findings of fact and conclusions of law.

The burden of proof is the structural fact that defines the proceeding. The agency must prove, by a preponderance of the evidence, that the cited standard applied to the cited condition, that the condition violated the standard’s terms, and that the employer knew or with reasonable diligence could have known of the condition. Employer knowledge is often the decisive issue: the agency must show the condition was known to the employer or knowable through reasonable diligence, which is why the inspection’s documentation of who knew what, and when, matters so much. An employer that can show it neither knew nor could reasonably have known of the violative condition has a defense that defeats the citation regardless of the condition’s danger. The knowledge element is the doctrinal recognition that the statute punishes culpable noncompliance, not the mere existence of hazards in a fallen world.

The employer’s defenses beyond lack of knowledge are the working tools of the employment bar. The employer may argue the standard did not apply to the condition cited, a question of regulatory interpretation the judge decides de novo. The employer may argue the agency’s reading of the standard is unreasonable or inconsistent with the standard’s text and history. The employer may argue that compliance was infeasible or would have created a greater hazard, the impossibility and greater-hazard defenses, which require the employer to show that no feasible means of compliance existed or that the available means would have endangered workers more than the cited condition. The employer may challenge the penalty’s calculation, the classification of the violation, and the abatement date. Each defense is litigated on the record the inspection created, which returns the practitioner to the first principle of the enforcement system: the inspection is the trial’s foundation, and a thin inspection record is a lost case for whichever side needed the missing evidence.

Commission review of the judge’s decision is the next stage, and its discretionary character surprises newcomers. The commission does not automatically review every judge’s decision. A party aggrieved by the decision may petition for review, and the commission decides whether to take the case, generally reserving its review for cases presenting important questions of law or policy, conflicting judge decisions, or significant penalty issues. When the commission does review, it decides the case on the record made before the judge, without new evidence, and issues a decision that becomes the commission’s final order. When it declines review, the judge’s decision becomes the commission’s final order by operation of law. Either way, the resulting order is the agency action that the court of appeals reviews under section 12.

The separation of the commission from the enforcement agency shapes the litigation culture in ways worth making explicit. Because the agency must prove its case to a neutral tribunal that owes it no deference as a litigant, the agency’s attorneys select cases with an eye to provability, settle cases where the record is weak, and build inspections with the trial in mind. Because employers face a prosecutor rather than a regulator, they invest in defense counsel, expert witnesses, and record challenges, and they settle where the record is strong. The result is a system in which most citations settle, a meaningful fraction are litigated to decision, and the litigated fraction disproportionately involves the important legal questions, which is how the case law develops. The commission’s docket is thus both the enforcement system’s courtroom and its lawmaking forum, and the independence that makes it a fair tribunal is the same independence that makes enforcement slow, expensive, and uncertain for both sides.

Penalties: What a Violation Costs

Section 17, codified at 29 U.S.C. 666, sets the civil penalty structure, and the numbers are best stated exactly, because they are the statute’s own. For a willful or repeated violation, the maximum penalty is $70,000 per violation, with a minimum of $5,000 for a willful violation. For a serious violation, the maximum is $7,000. For a violation determined not to be of a serious nature, an other-than-serious violation, the maximum is $7,000. For failure to correct a violation within the abatement period, the failure-to-abate penalty may reach $7,000 for each day the violation continues beyond the deadline. For failure to post the citation as required, the maximum is $7,000. These are the pre-adjustment levels set by section 3101 of Public Law 101-508 in 1990, which last raised the statutory maximums.

The classification determines the penalty, so the definitions matter. A serious violation exists where there is a substantial probability that death or serious physical harm could result from the hazard, unless the employer did not and could not with reasonable diligence have known of the violation. A willful violation is one committed with intentional disregard of the statute’s requirements or plain indifference to them, a state-of-mind finding that the agency must prove and that employers contest fiercely because the penalty is ten times the serious maximum and the label carries reputational weight. A repeated violation is one substantially similar to a prior violation for which the employer was previously cited, with the lookback reaching across the employer’s history. An other-than-serious violation is one that does not meet the serious threshold but still violates a standard. The failure-to-abate penalty, at $7,000 per day, is the provision with the most leverage in settlement negotiations, because the meter runs every day the hazard remains uncorrected after the deadline.

Three facts about penalties belong together, because the standing argument about the agency’s teeth takes each of them in isolation. First, the statutory maximums are modest relative to other regulatory regimes: environmental statutes routinely authorize penalties an order of magnitude larger per violation per day, and securities and financial statutes larger still. Second, the penalties that are actually assessed and collected are typically well below the statutory maximums, because the agency’s penalty calculation policies reduce the gravity-based starting figure for employer size, good faith, and history, and because settlements discount further. Third, the criminal provision is the most cited structural criticism in the field: a willful violation that causes a worker’s death is a misdemeanor, not a felony, with a short maximum sentence, which means the statute treats killing a worker through a willful safety violation more leniently than many regulatory schemes treat paperwork offenses. The pillar article carries the full treatment of the criminal provision and its history.

Presenting the three facts together is the honest way to address the claim that the agency is either toothless or overreaching, and the record supports neither characterization in the abstract. The rulemaking side is slow and constrained by the judicial doctrines this article has traced. The general duty clause gives the agency broad reach in individual cases, allowing citations where no specific standard addresses the hazard. And the penalties are modest relative to other regimes, which limits the deterrent punch of any individual citation. An account that reports only the slow rulemaking will conclude the agency is toothless; an account that reports only the general duty clause’s reach will conclude it is overreaching; an account that reports all three, with the penalty structure beside them, will conclude that the statute is a distinctive and somewhat odd instrument, powerful in its reach into the individual workplace and limited in the systemic force it can bring to bear. That is the complication the brief requires addressed, and the three facts together are the address.

The Abatement Clock

The citation’s most immediate command is not the penalty. It is the abatement date: the deadline by which the employer must correct the cited violation. The penalty punishes the past; the abatement date protects the future, and the statute treats the future as the more urgent matter. When the agency issues a citation under section 9, it states what must be fixed and by when, and the employer’s first obligation is to meet that date. An employer that fixes the hazard promptly but contests the penalty has understood the system’s priorities correctly.

The abatement obligation runs on its own clock, separate from the contest clock. The 15 working days to contest the citation do not automatically extend the abatement date, and an employer that assumes otherwise learns the lesson through the failure-to-abate penalty: $7,000 for each day the violation continues beyond the deadline. The daily meter is the provision with the most leverage in the enforcement system, because it converts delay from a litigation strategy into a compounding liability. An employer contesting the characterization of a violation while leaving the hazard in place watches the meter run every day, which is why experienced counsel advise clients to abate first and argue later. The hazard corrected is the workers protected; the legal argument about whether the citation was properly classified can proceed without anyone continuing to be exposed.

An employer that genuinely cannot meet the abatement date has a procedural path: it may seek modification of the abatement period through the commission, showing good-faith efforts to comply and the reasons more time is needed. The modification proceeding is the enforcement-side analogue of the temporary variance, and it serves the same function: adjusting a uniform deadline to the reality of a particular workplace without excusing noncompliance. Like the variance, it requires an evidentiary showing, and like the variance, it is used sparingly, because most abatement dates are set with achievable compliance in mind and most employers meet them.

The abatement system reveals the statute’s theory of enforcement more clearly than the penalty schedule does. The penalties are modest, as this article has shown, and no serious observer believes a $7,000 maximum deters a large employer on its own. What deters is the combination: the inspection that finds the hazard, the citation that names it publicly, the abatement date that requires it fixed, the daily meter that punishes delay, and the commission proceeding that tests whether the agency proved its case. The system is designed to produce corrected workplaces more than punished employers, and the abatement clock is the mechanism that keeps the design honest. A reader who understands the abatement obligation understands what enforcement is for, which is not primarily retribution for past violations but the removal of present dangers.

The General Duty Clause: Enforcement’s Second Engine

The general duty clause is the pillar article’s subject in full, but this article cannot map enforcement without describing the clause’s operational role, because the clause is the enforcement tool the agency reaches for when the rulebook runs out. Section 5(a)(1) of the act, codified at 29 U.S.C. 654(a)(1), requires each employer to furnish employment free from recognized hazards likely to cause death or serious physical harm. Where a specific standard addresses the hazard, the agency cites the standard. Where no standard addresses it, or where the standard is technologically dated in the way the borrowed 1971 rulebook often is, the agency cites the general duty clause. The clause is the catch-all that keeps enforcement from stopping at the edge of the standards.

Proving a general duty violation requires the agency to establish four elements, and each element is litigated. The hazard must have existed in the cited workplace. The hazard must have been recognized, either by the employer itself or as a hazard recognized in the employer’s industry. The hazard must have been likely to cause death or serious physical harm. And the hazard must have been feasibly abatable, meaning there existed a means of eliminating or materially reducing it that the employer could reasonably have implemented. The four elements together mean the clause is not a roving commission to punish every danger. The agency must prove recognition, likelihood of serious harm, and feasible abatement in the individual case, on the individual record, to the independent commission.

The operational significance of the clause is best understood against the rulemaking constraints this article has traced. New standards are rare because Benzene and Cotton Dust make them hard to write. The emergency power is unused because the courts made it unusable. The borrowed safety standards are dated because updating them requires the same slow rulemaking. In that environment, the general duty clause is the agency’s pressure valve: it allows enforcement to reach hazards the rulemaking system cannot timely address, without any rulemaking at all. An inspector who finds workers exposed to a hazard no standard covers does not need to wait for a multi-year rulemaking. The inspector documents the four elements, the agency cites the clause, and the commission decides whether the proof holds. The clause thus converts the rulemaking bottleneck from a complete enforcement stop into a detour, at the price of litigating the same four elements case by case instead of settling them once in a standard.

That price is real, and it defines the clause’s limits. A standard, once written, applies across an industry without re-proving the hazard in each workplace. A general duty citation must prove recognition, serious-harm likelihood, and feasible abatement anew in every case, before a judge who will examine the employer’s evidence that the hazard was not recognized or not feasibly abatable. The clause is therefore powerful in the individual case and weak as a systemic regulatory tool: it reaches everywhere and settles nothing. Employers experience it as unpredictability, since the duty is stated at a level of generality that requires case law to give it content, and the case law is made one contested citation at a time. Workers and their representatives experience it as the provision that keeps the statute alive in the long intervals between rulemakings. Both experiences are accurate, and the tension between them is the standing argument about whether the clause is a necessary safety net or an invitation to arbitrary enforcement.

The clause also interacts with the significant-risk and feasibility doctrines in a way that rewards attention. The clause’s four elements perform, in miniature and case by case, the functions that Benzene and Cotton Dust perform for rulemaking across industries. Recognition plus likelihood of serious harm is the individual-case analogue of the significant-risk threshold: the agency must show the danger is real and serious, not speculative. Feasible abatement is the individual-case analogue of the feasibility mandate: the employer must have been able to do something about it. The parallel is not accidental. It reflects a consistent legislative and judicial judgment that the government should prove danger and achievability before punishing, whether it punishes by rule or by citation. A practitioner who understands the parallel can move between the rulemaking cases and the enforcement cases without losing the thread, because the thread is the same: prove the risk, show it was feasible to address, and only then impose the duty.

The State Plan Alternative

Section 18, codified at 29 U.S.C. 667, offers states a choice that reshapes enforcement geography. A state may submit its own occupational safety and health plan to the Secretary of Labor, and the Secretary must approve it if the plan meets the statutory criteria, central among them that the state’s standards and enforcement are at least as effective as the federal program’s. An approved state plan displaces federal enforcement for the private-sector workplaces it covers, with the federal agency retaining authority over issues the plan does not cover and over federal workplaces. The state plan alternative is also the answer to one of the statute’s most consequential coverage gaps: the federal act does not protect state and local government employees, but a state operating an approved plan must extend its protections to its own public workforce.

The state plans matter to this article for two reasons. First, they multiply the enforcement actors: in a state-plan state, the inspector at the door works for the state, applies the state’s standards, and issues citations under state law, while the federal standards and the federal case law construing them supply the floor below which the state may not fall. A practitioner in such a state must know two rulebooks and two procedural systems, and must understand that the federal doctrines traced in this article, the significant-risk threshold, the feasibility requirement, the warrant rule, the independent adjudication, apply as the baseline the state program must meet or exceed. Second, the state plans are the field’s natural experiment: they allow comparison of enforcement approaches across jurisdictions operating under the same statutory floor, which is why the series gives the federal-state division its own treatment in the federal versus state labor protections guide. The comparison belongs there; the mechanism belongs here, because no map of enforcement is complete without the states that chose to run their own programs.

What the Vote Counts Reveal

The three decisions at the center of this article were all decided by close votes, and the alignments repay attention because they show a Court working through the statute’s meaning rather than applying settled doctrine. Benzene was 5 to 4, with the significant-risk holding carried by a plurality rather than a majority. Cotton Dust was 5 to 3, with Justice Powell taking no part. Barlow’s was 5 to 3, with Justice Brennan taking no part. In each case, a single vote the other way would have changed the law, and the narrowness of the margins explains why the doctrines have remained contested ground for decades.

The movement of individual justices between the cases is the most instructive part. Justice Stevens joined the Benzene plurality’s judgment imposing the significant-risk threshold, and then joined Justice Brennan’s Cotton Dust majority rejecting cost-benefit balancing. Justice Marshall dissented in Benzene, warning against the threshold, and then joined the Cotton Dust majority imposing the feasibility mandate. The two justices thus swapped sides between 1980 and 1981, which looks inconsistent only if the two cases asked the same question. They did not. Benzene asked what the agency must prove before regulating, and Stevens and Marshall disagreed about the answer. Cotton Dust asked what standard the agency must meet once it regulates, and they agreed. The pair of votes shows two justices distinguishing the threshold question from the stringency question with complete consistency, and it is a useful corrective to the habit of reading the two decisions as a single deregulatory or pro-regulatory bloc.

The non-participations matter as well. Justice Powell took no part in Cotton Dust, and Justice Brennan took no part in Barlow’s. In a Court of nine, an empty chair changes the arithmetic: a 5 to 3 decision with one justice absent is a narrower institutional statement than the same lineup with a full bench, and litigants in later cases have probed the resulting uncertainties. The Benzene plurality is the most probed of all, precisely because a plurality holding invites the argument that the significant-risk threshold rests on shakier precedential ground than a majority holding would. The argument has not prevailed; the threshold has governed every health rulemaking since. But the closeness of the votes is part of why the field’s arguments have never fully settled, and why each new rulemaking relitigates, in substance if not in form, the questions the Court divided over.

There is a final observation about the dissents, which in this field have had unusual afterlives. Justice Marshall’s Benzene dissent, joined by Justices Brennan, White, and Blackmun, stated the case for letting the agency regulate without the threshold, and its logic animates the labor-side argument to this day. Justice Rehnquist’s Cotton Dust dissent, joined by Chief Justice Burger, stated the case for reading cost-benefit balancing into the statute, and its logic animates the business-side argument to this day. Justice Stevens’s Barlow’s dissent, joined by Justices Blackmun and Rehnquist, stated the case for warrantless inspections, and its logic animates the argument that the warrant requirement hampers enforcement. In each instance, the losing side’s reasoning survived as the intellectual foundation of a standing political position. The decisions settled the law; the dissents supplied the vocabulary for continuing to argue about it. That is often how closely divided cases work, and it is worth naming because it explains why the same three decisions can be cited by all sides of every subsequent debate.

The Variance Safety Valve

Section 6(d), codified at 29 U.S.C. 655(d), provides the variance mechanism, and it deserves fuller treatment than the paragraph it received earlier because it is the statute’s answer to the rigidity of uniform national standards. A standard written for an entire industry will fit some workplaces badly: the required control may be physically impossible in a particular facility, or the employer may have developed an alternative method that protects workers as well or better. The variance gives such employers a lawful path other than violation, and its structure reflects the statute’s consistent demand for proof.

The statute provides two kinds. A temporary variance is available to an employer that cannot comply with a new standard by its effective date despite good-faith efforts, and that is taking steps to come into compliance while protecting employees in the interim. A permanent variance is available to an employer whose alternative methods provide working conditions as safe and healthful as the standard requires. In both cases the employer applies to the agency, employees must be notified and given an opportunity to participate, and the agency decides on a record. The temporary variance acknowledges the reality of compliance lead time: even a feasible standard may require equipment that must be ordered, installed, and tested, and the statute does not demand the impossible on day one. The permanent variance acknowledges the reality of innovation: an employer that invents a better way should not be punished for deviating from the prescribed way, provided the protection is equal or better.

In practice the variance is used sparingly, and the reasons are instructive. The application requires the employer to demonstrate, with evidence, either the good-faith path to compliance or the equivalent protection of the alternative, which is itself an evidentiary undertaking comparable in miniature to the rulemaking the standard went through. The employer must also litigate, in effect, against the agency’s interest in uniformity: every variance granted is a precedent that other employers will cite, so the agency examines applications with the systemic consequences in mind. And the employees’ right to participate means the application is not a private negotiation between the employer and the agency but a proceeding in which the workers whose safety is at stake have a voice. The combination of evidentiary burden, systemic caution, and worker participation makes the variance a real but narrow doorway.

The variance’s scarcity tells its own story about how employers actually experience the standards. Most employers comply, most of the time, with most standards, because compliance is cheaper than contest and because the standards, for all their datedness, generally describe sensible precautions. The employers who need variances are the edge cases: the facility whose architecture defeats the prescribed control, the process whose chemistry makes the standard method dangerous, the innovator whose new equipment outperforms the old rule. For those employers, the variance is the difference between lawful operation and violation, and the employment lawyer’s job is to recognize the edge case early enough to build the evidentiary showing. The variance is thus the individualized counterpart to the general standard, and the pair together express the statute’s theory: uniform rules for the industry, adjusted justice for the workplace that does not fit.

How to Read the Enforcement Record

The agency publishes enforcement data each year: inspections conducted, violations cited, penalties proposed and collected. These numbers are among the most cited and most misread figures in the field, and the neutrality of this article requires explaining how to read them rather than reciting them. The brief for this article instructs that inspection numbers, citation rates, and penalty levels be described institutionally, with sources and periods, and that instruction reflects a hard truth: the numbers measure enforcement activity, not safety outcomes, and activity moves with policy choices that change across administrations.

Enforcement intensity varies with administrations, and the variation is a standing political argument. An administration that prioritizes programmed inspections of high-hazard industries will produce higher inspection counts and more citations per inspection. An administration that prioritizes compliance assistance and cooperative programs will produce lower counts and describe the shift as smarter regulation. Business constituencies read rising numbers as overreach and falling numbers as relief. Labor constituencies read rising numbers as vigilance and falling numbers as neglect. The institutional fact underneath the argument is the one the statute created: a finite inspection force, a vast regulated universe, and a targeting system that rations attention by hazard, history, and complaint. Any change in the rationing formula changes the numbers, and the formula is a policy choice.

Three cautions belong with every citation of the figures. First, inspections are not a random sample of workplaces, so inspection-based violation rates do not describe the compliance rate of American industry. Programmed inspections target the most hazardous workplaces, complaint inspections target workplaces where someone reported a problem, and fatality investigations target workplaces where someone died. A violation rate drawn from that pool describes the pool, not the country. Second, penalties proposed are not penalties collected. The proposed figure reflects the gravity-based calculation at citation; the collected figure reflects settlements, contest reductions, and commission decisions, and it is typically much lower. Citing the proposed total as the agency’s enforcement output overstates the financial consequence by the distance between accusation and judgment. Third, none of these figures measures injuries prevented, which is the outcome the statute exists to produce. The injury and illness data come from a different system, the employer logs and the workers’ compensation records, and connecting enforcement activity to injury outcomes requires the kind of careful study that raw enforcement totals cannot supply.

The honest use of the enforcement record is therefore comparative and institutional rather than absolute. Compare inspection priorities across periods to see what the agency emphasized. Compare citation classifications to see whether the agency was finding serious violations or paperwork violations. Compare penalty proposals to collections to see how much of the initial figure survived adjudication. And always ask what changed in the targeting formula before concluding that a rising or falling number means workplaces got more or less safe. The numbers are real, the sources are public, and the periods are stated in the agency’s reports. What they prove is what the agency did, not what the workplaces are like, and the distinction is the beginning of wisdom in this field.

Why the Case Law Is About Health, Not Safety

A reader who has followed this article’s doctrinal sections will have noticed that every major case concerns a health standard or an inspection, and none concerns a safety standard’s substance. The pattern is not an accident of litigation fashion. It follows from the structure the startup shortcut created, and it explains why the significant-risk and feasibility doctrines govern toxic substances while the safety rulebook has generated almost no Supreme Court law.

The safety standards were borrowed, not made. The May 1971 adoption took the ANSI and NFPA consensus standards wholesale, without notice, comment, hearings, or a rulemaking record. A borrowed standard presents a challenger with very little to attack: there is no risk assessment to dispute, no feasibility analysis to second-guess, no statement of reasons to parse for gaps. The challenge would have to argue that the consensus standard itself was arbitrary, which is a difficult case to make against a standard the relevant industry wrote for itself. And because the safety standards govern observable physical hazards, falls, machines, electricity, trenching, the underlying danger is rarely in doubt. No one litigates whether an unguarded saw blade is dangerous. The safety standards have therefore lived a quiet legal life, cited in thousands of enforcement actions and construed in almost no Supreme Court opinions.

The health standards were built from scratch, through the full machinery of section 6(b), and every element of that machinery is litigable. The risk assessment can be challenged as insufficient to show significant risk. The feasibility analysis can be challenged as insufficient to show achievability. The statement of reasons can be challenged as inadequate. The substantial-evidence standard invites the challenge by directing the court to examine the whole record. And because health hazards are invisible, cumulative, and probabilistic, the underlying danger is always in doubt in the way a saw blade is not. Whether a given exposure level presents a significant risk of material health impairment is a question on which reasonable scientists disagree, which means it is a question on which litigation thrives. The health standards have therefore lived a loud legal life, with the two most important decisions in the field’s history both arising from health rulemakings.

The emergency standard record confirms the pattern. Of the nine emergency temporary standards, eight concerned toxic substances: asbestos twice, organophosphorous pesticides, 14 carcinogens, vinyl chloride, benzene, DBCP, acrylonitrile. Only one, diving operations in 1976, concerned a safety hazard in the physical sense. The emergency power was designed for grave dangers, and the grave dangers the agency perceived were overwhelmingly chemical. The courts’ responses, stays and vacaturs, thus shaped the law of emergency health regulation, not the law of safety standards. The diving standard’s stay is the exception that marks the boundary: even a safety-related emergency standard faced the same judicial skepticism, because the skepticism attaches to the emergency procedure rather than to the type of hazard.

The asymmetry has a practical consequence for the practitioner. The lawyer defending a safety citation works with a mature, stable body of standards whose meaning has been settled by decades of commission decisions, and the fights are about application: did the standard cover this condition, did the employer know, was the violation serious. The lawyer challenging or defending a health standard works with a smaller, harder-fought body of rules whose meaning is still being made, and the fights are about validity: did the agency prove significant risk, did it prove feasibility, does the record hold together. The two practices require different skills, different experts, and different expectations about what litigation can achieve. The safety practice is about the facts of the workplace. The health practice is about the science of the record. Understanding which practice a case belongs to is the first step in handling it well.

Prove the Risk, Then Ignore the Cost

The namable claim of this article deserves its own statement, because it is the sentence a reader should carry out of the doctrinal sections. The two foundational decisions impose a significant-risk threshold and then bar cost-benefit balancing, which is an unusual and demanding combination that explains both why new health standards are rare and why the ones that exist are stringent.

Consider how unusual the combination is against the background of American regulatory law. The standard model gives the agency a substantive goal and a balancing test: protect health, but only where the benefits justify the costs, or regulate where the agency finds the balance reasonable. That model produces a predictable pattern: frequent regulation, because the threshold for starting is low, with stringency moderated by the cost side of the balance. The workplace safety statute inverts the pattern. The threshold for starting is high, because Benzene requires a demonstrated significant risk before the agency may act at all. But once the threshold is crossed, the stringency is unmoderated by cost, because Cotton Dust requires the most protective feasible standard and forbids the agency from stopping where a cost-benefit analysis would have stopped. The result is the inverse pattern: infrequent regulation, each instance of it demanding.

The inversion has a logic, and stating it fairly requires acknowledging what each side of the field’s standing argument gets right. The business critique is that the combination produces rules whose costs exceed any reasonable measure of their benefits, imposed without the democratic check of a balancing test, and that the feasibility standard is cold comfort when feasibility is defined to tolerate very large costs short of industry destruction. The labor critique is that the combination is exactly what Congress intended: a statute that refuses to put a price on workers’ lungs, that treats the question how much protection as answered by what is achievable rather than by what is efficient, and that accepts slow rulemaking as the price of rules that actually protect. The Court in Cotton Dust chose the labor reading as a matter of statutory interpretation, holding that Congress wrote feasibility into the text and that the judiciary would not substitute balancing. The choice is law, and it has been law for more than three decades, but the argument about whether it is good policy continues, and an honest article reports the argument rather than pretending the law settled it.

The combination also explains the agency’s strategic behavior in a way that pure doctrine does not. An agency operating under prove-the-risk-then-ignore-the-cost faces a peculiar incentive structure. The expensive, slow, litigation-risky path is new health rulemaking, where the agency must build the risk assessment, survive the threshold fight, prove feasibility industry by industry, and then defend a stringent standard on substantial-evidence review. The cheaper, faster, litigation-tested paths are enforcement of existing standards, which requires no new rulemaking at all, and the general duty clause, which reaches hazards no standard addresses without any rulemaking. A rational agency allocates its marginal effort to the paths with the highest return per unit of legal risk, which means the doctrinal combination does not merely constrain rulemaking; it redirects the agency’s energy toward enforcement. The gap between statutory authority and delivered regulation is therefore not just a story about what the courts forbade. It is a story about what the courts made rational.

The Gap Between Authority and Delivered Regulation

The series thesis thread for this article is the gap between the authority Congress granted and the regulation actually delivered, with judicial doctrine setting the width of the gap. The 1970 statute reads, on its face, as a broad delegation: the agency shall set standards that most adequately assure safe and healthful working conditions, enforce them through inspections and citations, and punish violations with civil and criminal penalties. A reader coming to the text cold would expect a large and growing body of standards, frequently updated, backed by vigorous enforcement. The delivered reality is a rulebook whose founding safety layer was borrowed in 1971 and never fully rebuilt, a health standards program that slowed dramatically after the early 1980s, an emergency power unused for more than three decades, and an enforcement program that rations a finite inspection force across a vast regulated universe.

Every element of that gap traces to a judicial decision this article has examined. The significant-risk threshold of Benzene narrowed the doorway to health regulation. The feasibility mandate of Cotton Dust, while demanding in stringency, raised the evidentiary cost of walking through the doorway by requiring industry-by-industry proof of achievability. The warrant requirement of Barlow’s added a procedural step to every non-consensual inspection. The emergency standard record taught the agency that speed would be punished by stays and vacaturs. The independent commission guaranteed that enforcement would be litigated rather than imposed. None of these decisions rewrote the statute. Each of them construed it, and the construals compounded into a regulatory posture far more cautious than the text alone would suggest.

This is not a story of judicial sabotage, and it should not be told as one. Each decision has a defensible rationale rooted in values the legal system takes seriously: that the government should prove the danger before restricting liberty, that Congress’s words about feasibility mean what they say, that the Fourth Amendment protects the workplace as well as the home, that emergency powers need evidentiary discipline, that the prosecutor should not be the judge. A reader who finds those values attractive will find the decisions attractive, and will read the gap as the rule of law working as intended. A reader who counts the workers exposed while rulemakings grind through their second decade will read the same gap as a failure of the system to deliver what Congress promised. The article’s job is not to choose between those readings but to make the mechanism visible enough that the reader can choose with open eyes. The doctrine sets the width of the gap; the judgment about the width belongs to the reader.

For students and practitioners who must hold the whole machinery in mind at once, the discipline is to read the statute against the regulation and the cases together, rather than treating any one of them as the law. The statute states the authority, the regulation states the rule, and the cases state the conditions under which the rule survives. The series’ guide to how to read a federal statute works through that three-layer reading method in detail, and it is the right companion to this article for anyone moving from classroom study to practice. Readers preparing for examinations in administrative or employment law, where the standard-setting tests and the enforcement pathway are standard examinable material, often work through the pipeline below alongside the ReportMedic civics study page to fix each stage’s actor, test, and timeline in memory.

The Hazard-to-Citation Pipeline

Stage	Actor	Legal test	Timeline
Hazard identification	The agency and its research arm	Whether the evidence shows a hazard worth studying; for health standards, whether a significant risk exists	Years of study before rulemaking begins
Standard-setting	The agency through section 6(b)	Significant risk to regulate; most protective feasible standard; statement of reasons; substantial-evidence review	Federal Register notice, 30-day comment, hearing on request; routinely many years
Emergency standard	The agency through section 6(c)	Grave danger shown on an emergency record	Effective immediately; six-month maximum; 9 issued 1971 to 1983, none 1984 to 2015
Inspection	Compliance officers under section 8	Warrant or consent; administrative probable cause	Unannounced; targeted by program, complaint, or catastrophe
Citation	The agency under section 9	Violation of a standard or duty shown on the inspection record	Issued with reasonable promptness; proposed penalty by classification
Contest	Employer or employees under section 10; the independent commission	Whether the violation occurred and the penalty is appropriate; agency bears the burden	15 working days to contest; ALJ hearing then commission review
Appeal	Federal courts of appeals under section 12	Substantial evidence for facts; de novo for law	60 days to seek review

The pipeline is the article in miniature. A hazard enters at the top, carried by research and complaint. It becomes a rule only by surviving the significant-risk threshold, the feasibility mandate, and substantial-evidence review, a passage so demanding that the founding safety standards were borrowed rather than built and the emergency shortcut has sat unused for decades. The rule becomes a citation through an inspection that the Constitution requires be warranted or consensual. The citation becomes final only if uncontested within 15 working days; if contested, it passes to a commission entirely separate from the agency that issued it, and then to a court of appeals. At every stage, a different actor applies a different legal test on a different timeline, and the separation of the actors is deliberate. Congress built a system in which no single institution finds the hazard, writes the rule, inspects the workplace, and decides the contest, and the system’s virtues and its frustrations both flow from that refusal to concentrate power.

Frequently Asked Questions

Q: How does OSHA write a new safety standard?

Once the two-year consensus shortcut of section 6(a) expired, every new standard has traveled the ordinary road of section 6(b), codified at 29 U.S.C. 655(b). The agency publishes notice of the proposed rule in the Federal Register, interested persons have 30 days to submit written comments, any interested person may request a hearing with testimony and cross-examination, and the agency must publish the final standard with a statement of reasons. For health standards governing toxic substances, the agency must also make the significant-risk finding required by the Supreme Court’s 1980 Benzene decision and satisfy the feasibility mandate of the 1981 Cotton Dust decision, adopting the most protective standard that is technologically and economically achievable. The final rule is reviewable in a federal court of appeals under the substantial-evidence standard of section 655(f). In practice the process routinely takes many years, because the scientific record, the hearing, and the statement of reasons must all be built to survive that review.

Q: What is an OSHA emergency temporary standard?

An emergency temporary standard is a temporary rule the agency may issue under section 6(c), codified at 29 U.S.C. 655(c), when the Secretary finds that employees face a grave danger from a hazard and that an emergency standard is necessary to protect them. It takes effect immediately, without the notice, comment, and hearing procedures of section 6(b), serves as the proposed rule for the ordinary proceeding that must follow, and expires after six months unless replaced by a permanent standard. The judicial record of the power is poor: the agency issued exactly 9 emergency temporary standards between 1971 and 1983, and none from 1984 through March 2015, with most of the contested ones stayed or vacated by the courts. Because courts apply exacting review to the grave-danger finding on a necessarily thin emergency record, the agency has used the power very rarely for more than three decades.

Q: Can OSHA inspect a workplace without a warrant?

Not over the employer’s objection. In Marshall v. Barlow’s, Inc., 436 U.S. 307 (1978), the Supreme Court held 5 to 3 that the agency needs a warrant or the employer’s consent to inspect. The warrant uses an administrative probable cause standard, which the agency can satisfy with specific evidence of a violation, such as an employee complaint, or with a showing that the workplace was selected under reasonable administrative standards, such as a neutral inspection program. In practice most inspections proceed by consent: the compliance officer arrives unannounced, presents credentials, and the employer allows entry. The warrant requirement matters when an employer refuses, and it matters structurally, because it forces the agency to justify each non-consensual inspection in advance. An employer can challenge a defective inspection later before the review commission.

Q: How much can OSHA fine an employer?

Section 17 of the act, codified at 29 U.S.C. 666, sets the maximums, stated here at the pre-adjustment levels established by section 3101 of Public Law 101-508 in 1990. A willful or repeated violation carries up to $70,000 per violation, with a $5,000 minimum for willful violations. A serious violation carries up to $7,000. An other-than-serious violation carries up to $7,000. Failure to correct a violation by the abatement date can draw up to $7,000 for each day beyond the deadline. Failure to post the citation as required carries up to $7,000. Penalties actually assessed are typically well below these maximums, because the agency’s calculation policies reduce the starting figure for employer size, good faith, and history, and settlements discount further. The maximums are modest compared with the per-violation, per-day penalties available under many environmental and financial statutes.

Q: Must OSHA do a cost benefit analysis for standards?

No. In American Textile Mfrs. Inst. v. Donovan, 452 U.S. 490 (1981), the Supreme Court held 5 to 3 that the statute requires feasibility, not cost-benefit balancing. The agency must adopt the standard that most adequately assures worker health to the extent feasible, which means the most protective standard that is technologically and economically achievable. It may not stop short of that point because the remaining protection costs more than quantified benefits would justify, and it may not go beyond feasibility because benefits would justify it. Cost enters only through the feasibility door: a standard that would require nonexistent technology or destroy an industry’s economic viability is not feasible. This is a more demanding protection standard than most regulatory statutes impose, and it is one of the most commonly misunderstood points in the field.

Q: What is an OSHA state plan?

A state plan is a state-run occupational safety and health program approved under section 18 of the act, codified at 29 U.S.C. 667. A state may submit its own plan to the Secretary of Labor, who must approve it if the statutory criteria are met, centrally that the state’s standards and enforcement are at least as effective as the federal program’s. An approved plan displaces federal enforcement for the workplaces it covers. State plans matter for two practical reasons. First, the federal act does not cover state and local government employees, but a state with an approved plan must extend its protections to its own public workforce, closing the statute’s largest coverage gap. Second, employers in state-plan states answer to state inspectors applying state standards under state procedures, while the federal standards and the federal case law supply the floor below which the state program may not fall.

Q: How do you file an OSHA whistleblower complaint?

Section 11(c) of the act, codified at 29 U.S.C. 660(c), protects employees against retaliation for exercising rights under the statute, such as filing a safety complaint, participating in an inspection, or reporting an injury. An employee who believes they were discharged or otherwise discriminated against for such activity may file a complaint with the Secretary of Labor. The deadline is short: 30 days from the retaliatory action, which makes prompt action essential. If the Secretary finds merit, the Secretary may bring an action in federal district court seeking appropriate relief, which can include reinstatement and back pay. The complaint is filed with the agency, not with the review commission and not directly in court by the employee. Because the 30-day window closes quickly, workers who suspect retaliation should document the timeline and contact the agency without delay.

Q: Who decides a contested OSHA citation?

Not the agency that issued it. A contested citation goes to the Occupational Safety and Health Review Commission, created by section 11 of the act and codified at 29 U.S.C. 661. The commission is an independent adjudicatory body of three presidential appointees, entirely separate from the agency that writes standards and conducts inspections and entirely separate from the Department of Labor. The case is first heard by an administrative law judge of the commission, who takes evidence and issues a decision; the commission may then review that decision. The agency appears as a party bearing the burden of proving the violation. From the commission, section 12, codified at 29 U.S.C. 660, provides judicial review in the federal courts of appeals within 60 days. The recurring error is assuming the agency decides its own contests; Congress deliberately split prosecution from adjudication.

Q: Why does OSHA rulemaking take so many years?

The timeline is structural, not a sign of torpor. A health standard requires a risk assessment with exposure data and epidemiological evidence to satisfy the Benzene significant-risk threshold. It requires industry-by-industry technological and economic analysis to satisfy the Cotton Dust feasibility mandate. Section 6(b) then requires Federal Register notice, a 30-day comment period, and a trial-like hearing with testimony and cross-examination on request, followed by a statement of reasons answering the objections the hearing raised. The final rule faces substantial-evidence review in a court of appeals, which examines the whole record including contrary evidence. Each stage consumes staff time from the same scientists, engineers, and lawyers who also support enforcement, and the agency rationally over-builds the record as insurance against vacatur. Years are the price of a standard that survives.

Q: Why did Congress let OSHA adopt consensus standards without full rulemaking?

Congress faced a new agency with no scientific staff, no rulemaking record, and a mandate to produce enforceable standards immediately. Section 6(a) solved the startup problem by letting the Secretary adopt existing national consensus standards and established federal standards wholesale during a two-year window, without notice, comment, or hearings. On May 29, 1971, the Labor Department published a final rule at 36 FR 10466 adopting those standards en masse, effective August 1971, drawing principally on American National Standards Institute and National Fire Protection Association sources. The tradeoff was durability: voluntary industry guidance written in the 1960s became enforceable law overnight, and some of it remains technologically dated because the slow machinery of section 6(b) never got around to replacing it. The shortcut built most of the safety rulebook in a single day.

Q: What does substantial-evidence review mean for an OSHA standard in court?

It means the court of appeals examines whether the rulemaking record as a whole supports the standard, and sets the standard aside if it does not. Section 655(f) provides pre-enforcement review of section 6(b) standards directly in the courts of appeals, bypassing the review commission because a standard is a legislative act rather than an enforcement action against one employer. Substantial evidence is more than a scintilla but less than a preponderance: enough that a reasonable mind could accept it as adequate. The record considered as a whole includes the evidence cutting against the agency, so the agency must answer contrary evidence in its statement of reasons rather than merely pile up supporting studies. The standard is more demanding than the arbitrary-and-capricious review governing most agency rulemaking, and it is the reason rulemaking records are built, from the first study, with an appellate audience in mind.

Q: What counts as grave danger for an emergency standard?

Grave danger is the statutory trigger in section 6(c), codified at 29 U.S.C. 655(c), for an emergency temporary standard: the Secretary must find that employees face grave danger from exposure to a hazard before issuing a rule effective immediately without notice and comment. The finding must rest on evidence the agency can assemble quickly, and it is the natural target when the standard is challenged, because a court asked to stay an immediately effective rule scrutinizes the emergency justification first. The record shows how hard the finding is to defend: of the nine emergency standards issued from 1971 to 1983, most that drew litigation were stayed or vacated, and none were issued from 1984 through 2015. Grave danger therefore functions less as a routine tool and more as an extraordinary showing the agency must be prepared to defend on a thin, fast-built record. The six-month maximum duration adds a second constraint, since the agency must complete ordinary rulemaking for a permanent standard within that window.

Q: What is an OSHA variance and how does an employer obtain one?

A variance is permission to deviate from a standard, provided by section 6(d) of the act and codified at 29 U.S.C. 655(d). A temporary variance covers an employer that cannot comply with a new standard by its effective date despite good-faith efforts, while taking steps toward compliance and protecting employees in the interim. A permanent variance covers an employer whose alternative methods provide working conditions as safe and healthful as the standard requires. The employer applies to the agency, employees must be notified and may participate, and the agency decides on a record. Variances are used sparingly, because the application demands an evidentiary showing comparable in miniature to a rulemaking, and because each grant is a precedent other employers will cite. For the unusual workplace that genuinely does not fit the uniform rule, the variance is the lawful alternative to violation.

Q: What triggers an OSHA inspection?

Inspections are triggered by several routes. Employee complaints are a major source: a worker or representative may report a hazard, and the agency evaluates whether the complaint states reasonable grounds for an inspection while protecting the complainant’s confidentiality. Referrals from other agencies, media reports of serious incidents, and reports of fatalities or hospitalizations also trigger inspections, with the most severe events drawing the fastest response. Programmed inspections target high-hazard industries and workplaces under a neutral administrative plan, which the Supreme Court’s warrant decision expressly permits as a basis for inspection without individualized suspicion. Follow-up inspections check whether previously cited violations were abated. The mix of triggers means enforcement reaches workplaces the agency has never received a complaint about as well as those it has, and the neutral-plan route is what keeps the program from depending entirely on workers to report.

Q: How long does an employer have to contest an OSHA citation?

Fifteen working days from receipt of the citation, under section 10 of the Act codified at 29 U.S.C. 659. The employer notifies the Secretary of Labor of the intent to contest, specifying whether it challenges the citation itself, the proposed penalty, the abatement date, or some combination. The deadline is strict: missing it generally makes the citation a final order not subject to review, which is why counsel treat the receipt date as the start of an unforgiving clock. Employees and their authorized representatives have a parallel right to contest the reasonableness of the abatement period within the same fifteen working days. Once a timely contest is filed, the case goes to the Occupational Safety and Health Review Commission rather than being decided by the agency that issued the citation. Because the clock runs in working days rather than calendar days, weekends and federal holidays do not count, but no prudent employer cuts the calculation close.

Q: What is the difference between a commission administrative law judge and the review commission itself?

The administrative law judge is the trial-level adjudicator within the Occupational Safety and Health Review Commission system. The judge conducts the hearing, rules on evidence, hears witness testimony, makes factual findings, and issues an initial decision resolving the contested citation. The review commission itself is the three-member presidentially appointed body created by 29 U.S.C. 661 that sits above the judges. A party dissatisfied with a judge’s decision may petition the commission for review, and the commission may affirm, modify, reverse, or remand. Commission decisions establish the precedent that judges apply in later cases, so the body functions as both an appellate tribunal and the author of the system’s case law. From the commission, review goes to a federal court of appeals within sixty days. The two tiers matter for strategy because arguments must be preserved before the judge to be available on commission review.

Q: What is an abatement date and can it be contested?

The abatement date is the deadline the citation sets for correcting the cited violation, required by section 9 of the Act alongside the description of the violation and the proposed penalty. It tells the employer when the hazard must be fixed, and failure to meet it exposes the employer to additional penalties of up to seven thousand dollars per day under section 17. Yes, it can be contested: the employer’s fifteen-working-day contest under section 10 may challenge the citation, the penalty, the abatement date, or any combination, and employees have a parallel right to contest the reasonableness of the abatement period. Abatement-date fights often turn on feasibility and good faith, with employers arguing that engineering, procurement, or shutdown scheduling requires more time and the agency arguing that the hazard demands faster correction. Because the daily failure-to-abate penalty accumulates, the abatement date is frequently a central issue in settlement negotiations.

Q: What is the difference between a serious and a willful OSHA violation?

The difference is the employer’s state of mind and the penalty that follows. A serious violation exists where there is a substantial probability that death or serious physical harm could result from the hazard, unless the employer did not and could not with reasonable diligence have known of it. The maximum penalty is $7,000 per violation at the pre-adjustment levels. A willful violation is one committed with intentional disregard of the statute’s requirements or plain indifference to them, a state-of-mind finding the agency must prove and employers contest fiercely. The maximum penalty is $70,000 per violation, with a $5,000 minimum, ten times the serious maximum, and the willful label carries reputational weight beyond the fine. Classification drives penalty, settlement leverage, and litigation strategy, which is why the characterization of a violation is often contested as hard as its existence.

Q: What is a repeat OSHA violation and how far back does the agency look?

A repeat violation is a violation substantially similar to a prior violation that became a final order, meaning the earlier citation was not contested in time or was upheld through the contest process. The classification matters because repeat violations carry the same seventy thousand dollar maximum as willful violations, far above the seven thousand dollar cap for serious violations, and because a repeat finding signals to the commission and to the public that the employer failed to correct a known problem. The lookback question, how far into the past the agency may reach for the predicate violation, has been contested in commission and court decisions, with the agency historically applying a multi-year window and employers arguing for narrower limits. For compliance programs, the practical lesson is that every citation that becomes final becomes potential ammunition for a future repeat classification, which raises the stakes of deciding whether to contest.

Q: How does OSHA calculate the actual penalty amount?

The agency starts from the statutory maximums in 29 U.S.C. 666, which are seventy thousand dollars for willful or repeated violations and seven thousand dollars for serious, other-than-serious, posting, and per-day failure-to-abate violations, and then applies a penalty calculation system that adjusts within those caps. The gravity of the violation sets the base: how likely the hazard is to cause harm and how severe that harm would be. That base is then adjusted for the size of the employer, with smaller employers receiving reductions, for the employer’s history, with prior violations increasing the amount, and for good faith, with cooperative compliance efforts earning reductions. The classification chosen at citation, serious, willful, repeated, or other-than-serious, determines which cap applies and therefore dominates the outcome. Because the adjustments are structured rather than ad hoc, experienced practitioners can estimate the penalty range from the citation’s classification and the employer’s profile before negotiations begin.