A person who reads the enrolled text of Public Law 88-352 and stops there will get eight important things wrong. They will believe the employment title covers employers with twenty-five or more employees, that it exempts state and local governments and educational institutions, that the enforcement commission cannot go to court, that a successful plaintiff recovers no damages and has no right to a jury, that pregnancy discrimination falls outside the definition of sex discrimination, that the federal funding titles reach only the specific program that received the money, that the statute stops at the water’s edge, and that a pay claim expires on a clock that starts running when the employer sets the salary. Every one of those propositions was true of the act as signed. Not one of them describes the operative law. The Civil Rights Act of 1991 amendments are the largest single reason for the gap, and they are not the only reason.

This is the article that answers the question people actually mean when they ask whether the 1964 act is still in effect. It is, and asking whether a statute is in force is the easy half of the question. The useful half is what form it is in force in, because Congress has rewritten this one six times, four of those in direct response to Supreme Court decisions it disagreed with.

The pattern is regular enough to name, and this series calls it the override cycle: Title VII is the clearest example in American law of a statute maintained by legislative override, in which a judicial construction produces a legislative response that rewrites the text, so that the operative text cannot be read intelligently without knowing which decisions it was written against. Section 703(k) is unintelligible without Wards Cove Packing Co. v. Atonio. Section 703(m) is unintelligible without Price Waterhouse v. Hopkins. Section 1981a exists because the original statute had no damages. The definition of because of sex includes pregnancy because the Supreme Court held in 1976 that it did not. A reader who treats the codified text as a self-contained document is reading half of a conversation.

The statutory identities, stated once

The Civil Rights Act of 1964, Public Law 88-352, was signed on July 2, 1964, appears at 78 Stat. 241, and is codified principally at 42 U.S.C. sections 2000a and following, with the employment title at sections 2000e and following.

The Civil Rights Act of 1991, Public Law 102-166, was signed on November 21, 1991. It amends Title VII of the 1964 act, amends the Reconstruction-era contracts statute at 42 U.S.C. section 1981, adds a new damages provision at 42 U.S.C. section 1981a, and carries two additional titles of its own addressing advancement in the workplace and the employment rights of certain Senate and appointed employees.

The other amending statutes, in order: the Equal Employment Opportunity Act of 1972, Public Law 92-261, signed on March 24, 1972; the Pregnancy Discrimination Act of 1978, Public Law 95-555; the Civil Rights Restoration Act of 1987, Public Law 100-259, enacted on March 22, 1988 over a presidential veto; and the Lilly Ledbetter Fair Pay Act of 2009, Public Law 111-2, signed on January 29, 2009.

Public law numbers read as the Congress number followed by the sequential position of the law within that Congress, so Public Law 102-166 is the 166th law enacted by the 102nd Congress. The convention and its research uses belong to how to read a federal statute, which owns the citation mechanics for this series.

The original shape, and the three weaknesses that produced everything after

The act as passed had a specific architecture, walked through provision by provision in the title-by-title reading of the act. Three features of the employment title matter for the amendment history, and all three were the product of what the Senate would accept rather than of what the drafters preferred.

The first was the coverage threshold and its exclusions. Title VII as enacted reached employers with twenty-five or more employees, phased in over four years, and it excluded federal, state and local governments and educational institutions from the definition of employer. A statute that exempted every public school district, every state university and every municipal police and fire department was not going to reach a large share of the workplaces where the problem was most visible.

The second was the enforcement design. The Equal Employment Opportunity Commission received investigative and conciliation authority and no power to sue. When conciliation failed, the commission issued a notice and the aggrieved individual went looking for a lawyer. The reasons this design was chosen, and the trade that produced it during the Senate debate, are set out in how the act passed Congress, which owns the passage narrative. Its consequence is the subject of this article’s central claim.

The third was the remedy. A prevailing plaintiff could obtain an injunction, reinstatement, back pay and attorney’s fees, and nothing else. No compensatory damages for the harm actually suffered, no punitive damages regardless of the employer’s conduct, and, because the available relief was equitable, no right to a jury. For race claims a plaintiff could sometimes use the Reconstruction-era contracts statute instead, which carried no cap and a longer limitations period, and that alternative became important in ways nobody planned.

Those three weaknesses generated four of the six amendments. The fourth generator was the Supreme Court, and the fifth was a definitional question about the reach of the funding titles that arrived from an entirely different statute.

Wave one: the Equal Employment Opportunity Act of 1972

The first amendment was the largest structural change the employment title has received, and it is consistently underrated because it did not follow a famous Supreme Court decision. It followed seven years of the enforcement design not working.

By the end of the 1960s the commission had accumulated an inventory of unresolved charges it had no capacity to clear, because clearing a charge on which conciliation failed required a private lawyer to take the case. Congress had a full record of that failure by 1971, and the debate over the fix ran on a familiar axis: whether the commission should receive cease-and-desist authority, allowing it to order relief administratively in the manner of the National Labor Relations Board, or whether it should receive authority to sue in federal court. The administrative model was the stronger instrument and had passed the House in an earlier form. The court-enforcement model prevailed, and the compromise is why the commission litigates rather than adjudicates.

Public Law 92-261 made four changes that between them define the modern statute’s reach.

It lowered the coverage threshold from twenty-five employees to fifteen, which brought a large number of small and medium establishments inside the statute for the first time.

It removed the exclusions for state and local governments and for educational institutions, extending the title to public employers at every level below the federal, and to public and private schools, colleges and universities. That single change brought police departments, fire departments, school districts and state universities within reach of a statute that had exempted all of them, and it is the source of a substantial share of the litigation of the following two decades, particularly over written examinations and physical standards used in municipal hiring.

It gave the commission authority to bring civil actions in its own name when conciliation failed, transferring the cost of enforcement from the individual complainant to the government in the cases the agency chose to pursue. That is the single most consequential change in the entire amendment history if the measure is effect on how the statute operates, because it altered who bears the cost of making the prohibition real.

And it added a new section covering federal employment, establishing a distinct administrative process for federal employees with eventual access to federal court, a system that runs on different deadlines and through different offices than the private-sector process.

The 1972 statute also extended the charge-filing period and made adjustments to the deferral arrangements with state and local fair employment agencies. The mechanics of both belong to the Title VII charge process from filing to court, which owns the enforcement machinery.

Which employers came inside the statute in 1972?

It lowered the employer threshold from twenty-five employees to fifteen, brought state and local governments and educational institutions inside the definition of employer, added a separate process for federal employees, and gave the enforcement commission authority to sue in its own name when conciliation failed.

One further consequence deserves mention because it shows up in the research literature rather than in the case law. Because the extension applied to a defined set of newly covered employers on a known date, it created a comparison that economists have used to isolate the statute’s effect from everything else happening in the same period, a use discussed in what the act measurably changed. A coverage compromise negotiated in 1972 for political reasons became, decades later, one of the few credible identification strategies available in this field.

Wave two: the Pregnancy Discrimination Act of 1978

The second amendment is the cleanest single instance of the override cycle, and it took two years from decision to statute.

In General Electric Co. v. Gilbert, 429 U.S. 125 (1976), the Supreme Court held that an employee disability plan excluding pregnancy-related conditions from coverage did not discriminate because of sex under Title VII, reasoning that the plan divided employees into pregnant persons and non-pregnant persons rather than into women and men, and that the latter group includes members of both sexes. The holding was consistent with the Court’s earlier constitutional treatment of a state disability program and inconsistent with the enforcement commission’s own guidelines, which had taken the opposite position.

Congress responded by amending the definitions section rather than the operative prohibition, which is the more durable drafting choice. The Pregnancy Discrimination Act of 1978 provides that the terms because of sex and on the basis of sex include, but are not limited to, because of or on the basis of pregnancy, childbirth or related medical conditions, and that women affected by those conditions shall be treated the same for all employment-related purposes, including receipt of benefits under fringe benefit programs, as other persons not so affected but similar in their ability or inability to work.

Two features of that drafting matter. Amending a definition reaches every provision in which the defined term appears, so the change operated across hiring, discharge, compensation, terms and conditions, and benefit plans at once, without Congress having to enumerate them. And the comparison the statute specifies, to other persons similar in their ability or inability to work, sets up the analytical question that later litigation turned on, since an employer’s obligation depends on how it treats non-pregnant employees with comparable limitations rather than on any standalone accommodation duty.

The statute did not resolve everything it touched. Whether an employer that accommodates some categories of temporarily limited workers must accommodate pregnant workers on the same terms remained contested for decades, and the exclusion of abortion from the definition, with stated exceptions, was written into the amendment itself. But the core reversal was complete: the proposition that excluding pregnancy from a benefit program is not sex discrimination has not been law since October 1978.

Wave three: the Civil Rights Restoration Act of 1987

The third amendment did not begin with a Title VI case at all, which is why it is missing from most accounts of this statute’s evolution.

Grove City College v. Bell, 465 U.S. 555 (1984), construed the phrase program or activity in Title IX of the Education Amendments of 1972, a statute drafted on the model of Title VI and using its language. The Court held that where federal money reached an institution through student financial aid, the non-discrimination obligation attached to the specific program receiving the assistance rather than to the institution as a whole. Because the same phrase appears in Title VI of the 1964 act, in the disability provision of the Rehabilitation Act of 1973, and in the Age Discrimination Act of 1975, the reading threatened the coverage of four federal civil rights statutes simultaneously.

The legislative response took four years and failed twice before it passed, and it was contested on grounds that had little to do with the original decision, primarily over the reach of institution-wide coverage into religious organizations and over abortion-related services. The Civil Rights Restoration Act of 1987, Public Law 100-259, was enacted on March 22, 1988 after both chambers overrode a presidential veto. It added definitions to all four statutes providing that program or activity means all of the operations of a department, agency, corporation, partnership or other entity, or of a state or local government instrumentality, any part of which receives federal financial assistance.

The effect on the 1964 act is straightforward and large: a college that accepts federal money in any program is covered institution-wide, a hospital that participates in a federal payment program is covered institution-wide, and a state agency that receives federal funds for one function carries the obligation across the entity. Title VI’s leverage, described in the impact article as the mechanism behind the fastest institutional changes of the 1960s, was restored to its pre-1984 scope by this statute.

Two points of precision are worth keeping. This is the only entry in the amendment ledger produced by a decision under a different statute. And it is the only entry enacted over a veto rather than signed, which makes it the one place in this history where the override mechanism operated in its constitutional rather than its metaphorical sense; the procedure itself is set out in the veto and the override explained.

The vetoed predecessor: the Civil Rights Act of 1990

The 1991 statute cannot be understood without the bill that failed the year before, and the failure is a genuine dispute about a real question rather than a procedural footnote.

The 1989 Supreme Court term produced a cluster of employment decisions that narrowed doctrines the lower courts had been applying: Wards Cove Packing Co. v. Atonio on the allocation of burdens in disparate impact cases, Price Waterhouse v. Hopkins on mixed motives, Patterson v. McLean Credit Union on the reach of the contracts statute, Lorance v. AT&T Technologies on the clock for challenging a seniority system, and Martin v. Wilks on collateral challenges to consent decrees. Each holding is stated precisely, with its practical consequence, in the cases that shaped the act, which owns the litigation history.

A bill responding to those decisions moved through both chambers in 1990. Its central provision would have restored the employer’s burden of proving business necessity in disparate impact cases, and it would have added damages to the Title VII remedy for the first time.

President Bush vetoed it on October 22, 1990. The veto message argued that the bill’s standards, particularly the business necessity language and the damages provisions, would create pressure on employers to adopt hiring and promotion practices keyed to numerical outcomes in order to avoid litigation exposure, and it used the word quotas to describe that pressure. The message stated support for the principle of strengthening remedies against intentional discrimination and objected to the mechanism.

The bill’s sponsors, principally Senator Edward Kennedy in the Senate and Representative Augustus Hawkins in the House, rejected that characterization, arguing that the bill restored the burden allocation that had governed employment law from 1971 to 1989 without incident, that no employer had adopted quotas under the pre-1989 standard, and that a rule requiring an employer to justify a practice that excludes disproportionately is not a rule requiring the employer to hire by the numbers.

The Senate voted to override on October 24, 1990 and fell one vote short of the two-thirds required. Whether the veto message’s prediction about employer behavior was accurate is not a question the record settles, and both positions were argued in good faith by people who had spent careers on this statute. What can be stated without adjudicating it is that the disagreement was about the practical consequences of a burden of proof, that it produced a one-vote outcome, and that the compromise reached the following year adopted language on business necessity that both sides could describe as a restoration.

What was the disagreement that produced the 1990 veto?

The veto message argued that its burden-of-proof and damages provisions would pressure employers toward numerical hiring to avoid litigation exposure. The sponsors argued it restored the standard that had operated from 1971 to 1989 without producing that effect. The Senate override failed by one vote.

Wave four: the Civil Rights Act of 1991, provision by provision

Negotiations resumed in 1991, with Senator John Danforth playing the central role in assembling a version the administration would sign. Public Law 102-166 was signed on November 21, 1991, and it is the most substantial rewrite the employment title has received. Taking its operative provisions in order is the only way to see what it actually did, because the summary version, that it strengthened the 1964 act, is true of some sections, false of others, and unhelpful about the rest.

Section 101 amended the Reconstruction-era contracts statute to define the right to make and enforce contracts as including the making, performance, modification and termination of contracts and the enjoyment of all benefits, privileges, terms and conditions of the contractual relationship. That reversed Patterson and restored a route for race claims that carries no damages cap and a longer limitations period than Title VII.

Section 102 created a new damages provision, codified at 42 U.S.C. section 1981a. A complaining party who establishes intentional discrimination under Title VII may recover compensatory damages for future pecuniary losses, emotional pain, suffering, inconvenience, mental anguish and loss of enjoyment of life, and punitive damages where the employer acted with malice or reckless indifference to the individual’s federally protected rights. Punitive damages are not available against a government, government agency or political subdivision. Where damages are sought, either party may demand a jury trial. Damages are not available in disparate impact cases, which is the single most misunderstood limit in the statute.

The sums are capped, and the caps are the provision most frequently misstated. The combined total of compensatory and punitive damages for each complaining party may not exceed fifty thousand dollars for employers with more than fourteen and fewer than one hundred and one employees, one hundred thousand dollars for employers with more than one hundred and fewer than two hundred and one, two hundred thousand dollars for employers with more than two hundred and fewer than five hundred and one, and three hundred thousand dollars for employers with more than five hundred employees. Back pay is not counted against the cap, because it is an equitable make-whole remedy rather than damages. The caps carry no indexing provision and have not been raised by statute since enactment, so their real value has declined with the price level, a fact that is the strongest available answer to anyone who describes the 1991 act as a straightforward strengthening.

Section 105 addressed the disparate impact burden. It provides that an unlawful practice based on disparate impact is established when a complaining party demonstrates that a particular employment practice causes a disparate impact and the respondent fails to demonstrate that the practice is job related for the position in question and consistent with business necessity, or when the complaining party identifies an alternative practice and the respondent refuses to adopt it. It also provides that where the complaining party demonstrates that the elements of a decisionmaking process are not capable of separation for analysis, the process may be analyzed as one employment practice. The word demonstrates is defined in the statute to mean meeting the burdens of production and persuasion, which is the drafting that reverses Wards Cove on the allocation question while retaining a modified version of its specific-practice requirement.

Section 106 prohibited adjusting the scores of, using different cutoff scores for, or otherwise altering the results of employment-related tests on the basis of race, color, religion, sex or national origin, a practice that had been used in some public-sector selection systems.

Section 107 added the motivating-factor rule. An unlawful practice is established when a protected characteristic was a motivating factor for any employment practice, even though other factors also motivated the practice. A companion provision limits the remedy: if the respondent demonstrates it would have taken the same action in the absence of the impermissible factor, the court may grant declaratory relief, certain injunctive relief and attorney’s fees, and may not award damages or order reinstatement, hiring, promotion or payment. That is the adjustment to Price Waterhouse, and it is a genuine compromise rather than a reversal.

Section 108 limited collateral attacks on litigated or consent judgments by persons who had actual notice and an opportunity to object, or whose interests were adequately represented, which addressed Martin v. Wilks.

Section 109 extended Title VII and the disability employment statute to United States citizens employed abroad by American employers or by foreign companies controlled by them, with an exception where compliance would cause the employer to violate the law of the country in which the workplace is located. That reversed EEOC v. Arabian American Oil Co.

Section 112 provided that the limitations period for a challenge to a seniority system runs from the adoption of the system, from the individual becoming subject to it, or from the individual being injured by its application, which reversed Lorance.

Section 113 allowed recovery of expert fees as part of an attorney’s fee award, addressing a 1991 decision that had excluded them.

The statute also carried two additional titles: one establishing a commission to study advancement of women and minorities in the workplace, and one extending employment protections to certain Senate and appointed employees, an area later expanded by the Congressional Accountability Act of 1995, which applied Title VII and other workplace statutes to the legislative branch itself.

The amendment ledger

The artifact this article contributes is a single table giving each amending statute, the decision or gap it answered, and the operative change it made to the text. A reader who holds this table can answer almost any question about the difference between the act as passed and the law as it operates.

Year	Statute	Public law	What it responded to	Operative change to the text
1972	Equal Employment Opportunity Act	92-261	Seven years of an enforcement design that produced findings without remedies	Threshold cut from 25 to 15 employees; state and local governments and educational institutions brought within the definition of employer; commission given authority to sue; separate federal-employee process created
1978	Pregnancy Discrimination Act	95-555	General Electric Co. v. Gilbert (1976)	Definition of because of sex amended to include pregnancy, childbirth and related medical conditions, with a same-treatment standard keyed to ability or inability to work
1988	Civil Rights Restoration Act of 1987	100-259	Grove City College v. Bell (1984), a Title IX case construing shared language	Program or activity defined institution-wide across Title VI and three other statutes; enacted over a veto
1991	Civil Rights Act of 1991	102-166	Wards Cove, Price Waterhouse, Patterson, Lorance, Martin v. Wilks, Aramco, and the absence of damages	Disparate impact burden allocated to the employer and codified; motivating-factor rule with a remedy-limiting defense; compensatory and punitive damages with caps by employer size and a jury trial right; contracts statute extended to post-formation conduct; extraterritorial coverage; seniority clock reset; limits on collateral attacks; score adjustment prohibited; expert fees recoverable
2009	Lilly Ledbetter Fair Pay Act	111-2	Ledbetter v. Goodyear Tire and Rubber Co. (2007)	Discriminatory compensation decision treated as an unlawful practice when adopted, when an individual becomes subject to it, and each time compensation is paid pursuant to it

Two entries in that table did not follow a Supreme Court decision at all, and both are worth noticing. The 1972 statute responded to an enforcement failure that Congress could observe directly in the commission’s caseload, which is a different and rarer legislative trigger than a court ruling. And the damages provisions of the 1991 act filled a gap that had existed since 1964 rather than reversing anything, which is why the statute is a mixture of restoration and innovation rather than a simple correction.

Wave five: the Lilly Ledbetter Fair Pay Act of 2009

The most recent entry is the fastest override in the sequence and the narrowest in scope.

Ledbetter v. Goodyear Tire and Rubber Co., 550 U.S. 618 (2007), held by a vote of five to four that the charge-filing period for a pay discrimination claim runs from the discriminatory compensation decision, and that paychecks issued afterward under that decision do not restart it. The plaintiff had worked as a supervisor for nearly two decades before learning that her pay had been set below that of comparable men, and on the majority’s reading her claim had expired long before she could have known it existed.

The Lilly Ledbetter Fair Pay Act of 2009, Public Law 111-2, was the first bill signed in the 111th Congress, on January 29, 2009. It provides that an unlawful employment practice occurs with respect to discrimination in compensation when a discriminatory compensation decision or other practice is adopted, when an individual becomes subject to it, or when an individual is affected by its application, including each time wages, benefits or other compensation is paid. It applies parallel amendments to the age discrimination statute, the disability statute and the Rehabilitation Act, and it applies to claims pending on or after May 28, 2007, the day before the Ledbetter decision.

Three things the statute did not do are worth stating, because it is routinely credited with all of them. It did not create a general equal pay guarantee or change what counts as pay discrimination; it changed only when the clock starts. It did not remove the two-year limit on back pay recovery preceding the filing of a charge. And it did not address pay secrecy, disclosure obligations, or the burden of proving that a pay disparity is discriminatory. The statute’s own operation, and the questions it left to other law, are covered in the Lilly Ledbetter Fair Pay Act of 2009.

What the 1991 act did not do

The claim that the 1991 statute simply strengthened the 1964 act is the most common error in writing about this history, and correcting it requires naming four specific limits.

It capped the damages it created. Before November 1991 a Title VII plaintiff could recover no compensatory or punitive damages at all, so the caps are a ceiling on a new remedy rather than a reduction of an existing one, and that is the fair way to describe them. But the ceilings apply per complaining party regardless of the severity of the conduct or the duration of the harm, they exclude government employers from punitive damages entirely, and they carry no indexing mechanism. A cap fixed in nominal dollars declines in real value every year it is not adjusted, and none of the four tiers has been changed since enactment.

It withheld damages from disparate impact claims. The damages provision reaches intentional discrimination. A plaintiff who proves that a selection device excluded protected groups without justification obtains equitable relief and fees, not damages, which means the theory that Griggs established and the 1991 act codified remains the less remunerative of the two and is pleaded accordingly.

It compromised on mixed motives rather than reversing Price Waterhouse. Liability attaches on a motivating-factor showing, which is more favorable to plaintiffs than the pre-1991 rule, and the employer’s same-decision showing cuts off damages, reinstatement and back pay, which limits the value of that liability substantially. Both halves are the statute.

And it retained the specific-practice requirement from the decision it was written against. A plaintiff must still identify the particular practice causing the disparate impact, with relief from that requirement available only where the elements of the process cannot be separated for analysis. Whether that exception is broad or narrow depends on how a given court reads the phrase, and the variation is real.

There is also a provision in the statute that reads oddly to anyone encountering it for the first time: the act includes language excluding certain parties in the litigation that produced Wards Cove from the new rules. Congress reversing a decision while exempting the case that produced it is unusual, and it is the kind of detail that a reader checking the text will find and that summaries never mention.

The retroactivity question and why 1991 cases split

One question dominated employment litigation for three years after the 1991 statute and is worth understanding because it recurs whenever Congress amends a remedial statute.

If a plaintiff’s claim arose in 1989 but had not been resolved by 1992, did the new damages provisions apply? The statute’s own effective-date language was ambiguous, deliberately so, because the drafters could not agree.

The Supreme Court answered in two decisions issued on the same day in 1994. Landgraf v. USI Film Products, 511 U.S. 244 (1994), held that the damages and jury trial provisions do not apply to conduct occurring before enactment, setting out a general framework: a court asks whether Congress has clearly prescribed the statute’s temporal reach, and if not, whether applying it would impair rights, increase liability for past conduct, or impose new duties, in which case the presumption against retroactivity applies. Rivers v. Roadway Express, 511 U.S. 298 (1994), applied the same reasoning to the amendment of the contracts statute.

The practical consequence was that claims arising before November 21, 1991 were litigated under the pre-1991 rules for years afterward, which is why reported decisions from the mid-1990s can apply Wards Cove and Patterson without being wrong. A researcher reading a 1995 opinion that applies a superseded standard is usually looking at a retroactivity result rather than an error.

The contrast with the 2009 statute is instructive. Congress had watched the retroactivity fight and wrote an explicit application date into the Ledbetter Act, which eliminated the question entirely. That is legislative drafting learning from litigation, and it is a small but real illustration of the override cycle producing better statutes over time.

The titles that were never amended

Six amendments in forty-five years is a great deal of legislative attention, and almost all of it went to one title. That distribution is itself informative.

Title II, the public accommodations title, has not been amended in any substantive way since enactment. Its covered categories are the ones Congress wrote in 1964, its exemption for private clubs and for owner-occupied lodging with five or fewer rooms for rent is unchanged, and its remedy remains injunctive relief plus attorney’s fees with no damages. Proposals to expand the covered categories or to add damages have been introduced across the decades without passing. The reason the title has attracted no legislative correction is the same reason it produced little litigation: the Supreme Court upheld it immediately and then had few occasions to construe it narrowly, so there was nothing for Congress to reverse.

Title III, on desegregation of public facilities, and Title IV, on public education, have likewise not been substantially amended. Title IV’s authorization for the Attorney General to sue and for technical assistance to districts operates as written.

Title VI was amended once, by the 1988 restoration act, and only as to the definition of program or activity. Its operative prohibition, its fund-termination mechanism and its agency-by-agency regulatory design are the 1964 text.

Title VII has been amended five times, in 1972, 1978, 1991, 2009 and, in a limited way, by the coverage extensions to legislative-branch employees. The disproportion is not evidence that Congress cared more about employment than about public accommodations. It is evidence that the employment title generated the most litigation, that litigation generates constructions, and that constructions are what Congress responds to. A statute that is litigated heavily will be amended more often than one that is not, independent of anything about its subject.

The statutes that changed Title VII without amending it

Several laws altered how the employment title operates without touching its text, and a reader tracing the operative law needs them.

The Age Discrimination in Employment Act of 1967 and the employment title of the Americans with Disabilities Act of 1990 are separate statutes that borrow Title VII’s charge-filing machinery and much of its vocabulary. Their causation standards differ, which produces recurring confusion when a decision under one is cited as authority for another, and the age statute in particular requires but-for causation on a reading the Supreme Court adopted in 2009 rather than the motivating-factor standard the 1991 act wrote into Title VII.

The Civil Service Reform Act of 1978 and the reorganization that accompanied it consolidated federal-sector equal employment functions under the commission, which changed the administration of the federal-employee process the 1972 statute had created.

The Congressional Accountability Act of 1995 applied Title VII and ten other workplace statutes to the legislative branch, creating a separate office and procedure for congressional employees. It is not an amendment to the 1964 act, and it is the reason the answer to whether Title VII covers congressional staff changed in 1995 without a word of Title VII changing.

The Genetic Information Nondiscrimination Act of 2008 added a protected category through a separate statute rather than by amending the definitions section, which is the drafting choice Congress did not make in 1978. Comparing the two approaches is instructive: amending a definition, as the pregnancy statute did, integrates the new category into every provision automatically, while a standalone statute requires its own enforcement provisions, its own exceptions and its own body of construction.

The Religious Freedom Restoration Act of 1993 supplies a framework that can interact with Title VII claims against certain defendants, an intersection the Supreme Court expressly declined to address in its 2020 decision on sexual orientation and gender identity.

And the arbitration statutes, though they say nothing about civil rights, have reshaped where Title VII claims are heard, since decisions construing the Federal Arbitration Act determine the enforceability of agreements routing employment claims out of federal court. That is a substantial change in the statute’s practical operation produced entirely outside it.

Why a construction is not an amendment

The 2020 decision holding that discrimination because of sexual orientation or gender identity is discrimination because of sex is routinely described as having amended Title VII or added categories to it. It did neither, and the distinction is not pedantic in an article about amendments.

Congress amends a statute by enacting text through both chambers and presentment. Every entry in the ledger above went through that process, and each one changed words on the page: the definition of because of sex acquired a clause in 1978, a new damages section entered the code in 1991, a new subsection on compensation accrual entered it in 2009. A person can read the difference.

A court construes a statute by determining what the enacted words mean. The 2020 decision reasoned from the text’s focus on discrimination against an individual because of that individual’s sex and from but-for causation, and the majority stated explicitly that the legislators who adopted the provision likely did not anticipate the application. Nothing in the United States Code changed. The holding states what the words have meant since 1964.

Three consequences follow, and they are why the distinction matters. A construction can be revisited by the courts in a way an enacted text cannot. A construction can be overridden by ordinary legislation, which is the entire subject of this article, and the fact that no such override has been enacted is a fact about legislative outcomes rather than about the decision’s status. And a person tracing the statute’s history through the Statutes at Large will find the six amendments and will not find the construction, which means research that relies only on the enrolled texts will miss a substantial part of the operative law. The full sequence of constructions, including which ones Congress reversed, is traced in the cases that shaped the act.

The amendments that did not happen

An honest evolution article names the proposals that failed, because the shape of the operative law is determined as much by what Congress declined to enact as by what it passed.

Cease-and-desist authority for the enforcement commission was the stronger of the two enforcement models debated in 1971 and 1972 and did not pass. Had it passed, the commission would order relief administratively and employers would seek review in court, which is the structure of federal labor law. The entire architecture of Title VII litigation, including the volume of private suits and the role of the courts in building the proof frameworks, follows from the choice Congress made instead.

Indexing of the damages caps has been proposed repeatedly since 1991 and has not been enacted. The practical effect of that inaction accumulates every year, and it is the clearest example in this history of a legislative default producing a substantive change without any vote.

Expansion of Title II’s covered categories, and the addition of damages to that title, have been proposed and have not passed, which is why federal public accommodations coverage remains narrower than the public accommodations law of many states.

Extension of the disparate impact theory to the funding titles by statute, reversing the 2001 decision that eliminated private enforcement of the effects regulations under Title VI, has been introduced without passing. That is the one significant judicial narrowing in this statute’s history that Congress has not answered, and its persistence is the best evidence that the override cycle is not automatic. It requires an organized constituency, a legible harm and an available legislative vehicle, and where any of the three is missing the construction stands.

The 1991 act’s other two titles

Public Law 102-166 is usually discussed as though it consisted only of the employment amendments, and it did not. Two further titles carry their own content.

The second title established a commission to study the artificial barriers to advancement of women and minorities into management and decisionmaking positions in business, and to make recommendations. It authorized study and reporting rather than any enforceable obligation, which places it in a familiar category: a legislative response that documents a problem without creating a duty. Its output entered the policy record and its recommendations were not enacted as requirements.

The third title extended workplace protections to certain Senate employees and to appointed officials at state and local level who had been outside the ordinary coverage, establishing procedures within the legislative branch rather than routing claims through the enforcement commission. The broader project of applying workplace law to Congress itself was completed four years later by the Congressional Accountability Act of 1995, which covered legislative-branch employees under Title VII and ten other statutes through an office created for the purpose.

Both titles illustrate something about how a large civil rights measure assembles. A bill that must gather votes across two years and survive a veto accumulates provisions that individual members need, and the enacted statute is broader in subject and thinner in obligation than the core provisions suggest. Reading only the sections that changed section 2000e-2 gives an accurate picture of the operative change and an incomplete picture of the legislation.

How the amendments changed litigation practice

Statutory text changes what lawyers do, and the three biggest changes in this history are visible in practice rather than in doctrine.

The 1972 grant of litigation authority created a public enforcement track alongside the private one. The commission could select cases with systemic significance, fund them, and pursue pattern-or-practice theories that an individual plaintiff would not have the resources to develop. That capacity is the reason the systemic consent decrees of the 1970s exist, and those decrees account for a meaningful share of the occupational change the economic literature detects.

The 1991 creation of damages and jury trials changed the economics of the private track completely. A claim worth back pay and an injunction is worth what the lost wages are worth, and a claim carrying compensatory and punitive damages with a jury deciding them is worth something else, including a settlement value driven by outcome variance rather than by computable loss. The plaintiffs’ employment bar expanded after 1991 because the cases became financeable. Employment practices liability insurance grew into a distinct product line for the same reason. And employers’ incentive to invest in documentation, training and internal complaint procedures rose, because those measures reduce exposure to a jury.

The 1991 shift of the impact burden and the 2009 accrual change were narrower in their effect on practice, but both altered pleading. After 1991 a plaintiff must plead a particular practice, which requires knowing something about the employer’s process before discovery. After 2009 a pay claim survives that would previously have been dismissed on the pleadings, which moved a category of case past the earliest stage and into discovery, where compensation records exist.

A reader who wants the procedural sequence in detail, from the charge through conciliation to suit, will find it in the Title VII charge process from filing to court.

The interpretive questions the 1991 act created

An override statute answers old questions and creates new ones, and the 1991 act created four that occupied the courts for years afterward.

The first was retroactivity, resolved in 1994 and described above.

The second was the meaning of business necessity. The statute requires the employer to demonstrate that a practice is job related for the position in question and consistent with business necessity, and it contains an interpretive-memorandum provision limiting the legislative history that may be used to construe that phrase, which is an unusual drafting device and a direct product of the negotiation. The result is that courts construing one of the statute’s central terms are working with a restricted record by statutory command.

The third was the standard for punitive damages. The statute requires malice or reckless indifference to federally protected rights, and the Supreme Court addressed what that requires in Kolstad v. American Dental Association in 1999, holding that the standard focuses on the employer’s state of mind regarding the legality of its conduct rather than on the egregiousness of the conduct, and that an employer may avoid vicarious punitive liability for a manager’s decision where it has made good-faith efforts to comply. That holding tied the damages provision to the same compliance-incentive logic the harassment cases had produced a year earlier.

The fourth was the relationship between the motivating-factor provision and the rest of the statute. Because section 107 lists the protected characteristics and does not mention retaliation, the Supreme Court held in 2013 that retaliation claims require but-for causation rather than the motivating-factor standard. That is an override statute’s language producing a distinction Congress did not discuss, which is a common consequence of legislating against specific decisions rather than restating a framework whole.

What the evolution cost, and what it shows about legislating

Each amendment in the ledger carried a political price, and naming the prices makes the pattern legible.

The 1972 statute cost the stronger enforcement model. Cease-and-desist authority was the price of passage, and the court-enforcement compromise has shaped every subsequent decade of this statute’s operation.

The 1978 statute was comparatively cheap, passing with broad support, which is what an override looks like when the harm is concrete and the reversal is narrow.

The 1988 statute cost four years and required a veto override, and it acquired provisions on unrelated subjects during the delay, which is what happens to a bill that must be reassembled across Congresses.

The 1991 statute cost the caps, the exclusion of impact claims from damages, and the same-decision limitation. It also cost a year, a veto and a one-vote override failure. A reader assessing whether the 1991 act was a victory for its sponsors should hold all of that together: they obtained a codified impact standard, damages and juries, and they paid for them with ceilings that have never been raised.

The 2009 statute cost almost nothing procedurally, passing early in a Congress with an aligned President, which is the condition under which an override moves fastest.

The general lesson generalizes across this series. Overrides are cheapest when the decision being reversed produced a concrete, legible harm, when the reversal can be drafted narrowly, and when the legislative calendar and the veto point are favorable. When any of those fails, the override either does not happen or arrives with the price written into the text, and the price is usually a limit that outlives the memory of the negotiation that produced it.

The enforcement-model debate of 1971 and 1972

The choice between administrative adjudication and court enforcement is the most consequential decision in this amendment history, and it is worth setting out because the argument recurs whenever Congress builds an enforcement agency.

The case for cease-and-desist authority rested on speed, cost and consistency. An agency that can order relief resolves a matter without the complainant retaining counsel, without a filing fee, and without waiting for a court’s calendar. It develops a body of consistent decisions because the same institution decides every case, which reduces the variation that comes from hundreds of district judges applying an unfamiliar statute. And it shifts the cost of enforcement decisively onto the government, which is what an aggrieved worker without resources needs. Supporters pointed at federal labor law, where the National Labor Relations Board had operated on that model since 1935, as proof the design worked at scale.

The case against it rested on due process, on the concentration of power, and on the character of the decision being made. Opponents argued that an agency that investigates, prosecutes and adjudicates the same charge combines functions that should be separated, that discrimination findings can impose serious reputational and financial consequences without a jury or an independent factfinder, and that an agency committed institutionally to the statute’s purposes is not the right body to weigh a contested factual dispute between an employee and an employer. Some also argued that the labor board comparison cut the other way, pointing at that agency’s record of doctrinal swings across administrations.

Congress chose court enforcement, giving the commission authority to file civil actions rather than to order relief. The consequences run through everything in this article. Because the courts decide, the proof frameworks were built judicially rather than administratively, which is why McDonnell Douglas and Griggs are Supreme Court decisions rather than agency rules. Because the courts decide, the doctrines could be narrowed by the Supreme Court in 1989, which is what produced the 1991 override. And because the courts decide, a private plaintiff’s access to counsel remained central, which is why the addition of damages in 1991 changed the statute’s practical reach as much as any doctrinal provision did.

A reader assessing whether Congress chose correctly should notice that the question is still contested and that both models have produced records that supporters of the other can point to. What is not contested is that the choice was determinative, which is the general point: the enforcement model is not an implementation detail attached to a prohibition, it is the thing that decides what the prohibition amounts to.

How the 1991 negotiation worked

The path from a vetoed bill in October 1990 to a signed statute in November 1991 involved a specific set of moves, and understanding them explains why the enacted text reads the way it does.

The core disagreement was narrow and technical: whether an employer defending a practice with a disparate impact bears a burden of persuasion or only a burden of production, and how demanding the justification standard is. Everything else, including damages, was negotiable around that question, and the sponsors’ willingness to accept caps on damages was the currency that eventually bought agreement on the burden.

Senator John Danforth acted as the principal broker on the Republican side, working with the sponsors of the earlier bill and with the administration to find language on business necessity that both could describe favorably to their own constituencies. The result is the phrase job related for the position in question and consistent with business necessity, which appears in the statute without a definition, accompanied by an unusual provision limiting the legislative history that may be consulted in construing it. That device exists because the parties could agree on the words and not on what they meant, and rather than leave conflicting floor statements to be mined by litigants, the statute restricts what may be used. It is a rare instance of Congress legislating about its own legislative history, and it tells a reader more about the negotiation than any single provision.

The damages caps came from the same trade. The administration’s stated objection to unlimited damages was that exposure without a ceiling would drive settlement behavior regardless of merit, and the tiered structure by employer size was the compromise. The sponsors accepted it to obtain the burden allocation and the jury trial right. Nobody proposed indexing at the time, which is the kind of omission that looks minor in the year of enactment and compounds indefinitely.

The other decisions being reversed were less contested and moved as a package. Restoring the contracts statute to post-formation conduct, extending coverage abroad, resetting the seniority clock and limiting collateral attacks on consent decrees each had constituencies and none carried the ideological weight of the burden question.

A reader who wants the procedural anatomy of how a bill assembles and survives a veto point will find it in how a bill becomes law in practice, which owns the process explanation for this series.

Each 1991 section, applied to a case

Statutory sections are easier to hold when attached to the situation they govern, so here is the 1991 act read through a single hypothetical charge.

A worker is passed over for promotion. She believes sex was a factor, and she also believes the promotional examination the employer used screens out women at a higher rate.

Her intentional discrimination claim is governed by section 107. If she demonstrates that sex was a motivating factor in the decision, liability attaches even if other factors also motivated it. If the employer proves it would have promoted the other candidate anyway, she obtains declaratory relief, certain injunctive relief and attorney’s fees, and not damages, reinstatement or promotion. Before 1991 the same showing by the employer would have defeated liability entirely.

Her impact claim is governed by section 105. She must identify the examination as the particular practice causing the disparity, unless she can show that the elements of the promotion process are not capable of separation for analysis, in which case she may challenge the process as one practice. The employer must then demonstrate, meeting both the burden of production and the burden of persuasion, that the examination is job related for the position and consistent with business necessity. Before 1991 the employer would have carried a burden of production only.

Her remedy is governed by section 102. On the intentional claim, if she prevails without the employer establishing the same-decision defense, she may recover back pay outside any cap, and compensatory and punitive damages within the cap for her employer’s size, with a jury deciding if either party demands one. Punitive damages require malice or reckless indifference, and are unavailable if the employer is a government entity. On the impact claim she recovers equitable relief and fees, and no damages regardless of how the examination performed.

If she is a United States citizen working for the employer’s foreign subsidiary, section 109 brings her within the statute unless compliance would require the employer to violate local law.

If the employer had adjusted examination scores by sex to avoid the disparity, section 106 would prohibit that, which is the provision that makes the tension between the two theories acute and that the Supreme Court addressed in 2009 with a strong-basis-in-evidence standard.

Running an actual set of facts through the sections is the fastest way to see that the 1991 act is not one policy but six, and that they do not all point the same direction.

This statute’s amendment record against the rest of the series

Placing the 1964 act’s evolution next to other major statutes clarifies what is distinctive about it.

Some statutes evolve by reauthorization on a schedule. The Voting Rights Act of 1965 carried expiring provisions that forced Congress to revisit it repeatedly, which produced a different kind of amendment history: periodic, calendar-driven, and focused on extending or modifying a coverage formula rather than on reversing constructions. That record is traced in the reauthorizations article of the voting cluster.

Some statutes evolve through the appropriations and budget process, with substantive changes riding on must-pass vehicles, which produces amendments that are hard to trace because they are not enacted as freestanding civil rights measures.

Some statutes barely evolve at all, either because they generate little litigation or because no coalition exists to revisit them.

The 1964 act belongs to a fourth category: amendment driven by judicial construction, on no schedule, with each amendment written against specific decisions. That produces a text full of provisions that read as answers to questions the statute does not ask, and it produces a research requirement that is unusual. To read the Voting Rights Act’s evolution a researcher follows the reauthorizations. To read this one, a researcher has to follow the case law and the amendments together, because half the enacted language is a response to the other half of the story.

The general lesson is that a statute’s amendment pattern is a function of its enforcement design. A statute enforced through litigation will accumulate constructions and therefore overrides. A statute enforced administratively with sunset provisions will accumulate reauthorizations. Neither pattern is evidence of more or less legislative attention; they are different shapes of attention produced by different machinery.

The codified architecture of the operative statute

A reader who wants to work with the operative text rather than with a narrative of how it got there needs to know where the pieces live, because one act does not sit in one place in the code.

The employment title is codified at 42 U.S.C. sections 2000e and following. Section 2000e carries the definitions, including the fifteen-employee threshold, the extension of because of sex to pregnancy, and the definition of religion to include observance and practice subject to accommodation without undue hardship. Section 2000e-2 carries the operative prohibitions, with the disparate impact provision at subsection (k) and the motivating-factor rule at subsection (m), both of which entered the text in 1991. Section 2000e-3 carries the retaliation prohibition. Section 2000e-4 establishes the commission, and section 2000e-5 carries the charge and enforcement procedure, including the filing deadlines, the deferral arrangements, the commission’s litigation authority added in 1972, and the compensation-accrual provision added in 2009. Section 2000e-16 carries the federal-employee provisions added in 1972.

The damages provision does not live with the rest of the title. It is at 42 U.S.C. section 1981a, placed next to the Reconstruction-era contracts statute at section 1981 rather than in the Title VII chapter, which is a codification choice that catches researchers out regularly. A person searching the 2000e sections for the damages caps will not find them.

The public accommodations title is at sections 2000a and following, the public facilities title at 2000b, the education title at 2000c, and the federally assisted programs title at 2000d and following, with the program-or-activity definition added in 1988 at section 2000d-4a.

Two research habits follow from this scatter. Always check whether a section was in the original act or was added later, because a section number in the 2000e range gives no indication of its vintage. And when tracing a provision back, move from the code section to the credits that identify the public law, then to the Statutes at Large text of the amending act, since the code’s editorial arrangement can obscure what a single amending statute did across several sections at once. The technique belongs to how to find legislative history and to the difference between the US Code and the Statutes at Large, which own the research mechanics for this series.

What the amending statutes say about themselves

Several of these statutes carry findings and purposes sections, which are unusual in the 1964 act itself and increasingly common afterward, and they are useful evidence about what Congress thought it was doing.

The 1964 act contains no findings section of the kind later civil rights statutes carry. Its coverage provisions perform that function implicitly by tying each category to a commerce connection, and the evidentiary record supporting the commerce theory sits in the committee reports and the hearings rather than in the enacted text.

The 1991 statute is different. It opens with findings and purposes stating that legislation is necessary to provide additional protections against unlawful discrimination in employment, that the decisions of the Supreme Court had weakened the scope and effectiveness of federal civil rights protections, and that its purposes include codifying the concepts of business necessity and job relatedness enunciated in Griggs v. Duke Power Co. and in the decisions prior to Wards Cove, responding to recent decisions by expanding the scope of relevant civil rights statutes, and confirming statutory authority and providing statutory guidelines for the adjudication of disparate impact suits.

Two things follow. The statute names the decisions it is written against, in its own text, which removes any doubt about the override character of the enactment and which is why an article describing this history can be precise rather than inferential. And it names Griggs as the standard being restored, which is a drafting instruction to courts construing the business necessity phrase and which sits alongside the provision limiting other legislative history. A court asked what business necessity means has, by statutory command, a named case to look at and a restricted record otherwise.

The 2009 statute follows the same model, opening with findings on the concealment of pay disparities and stating that the Ledbetter decision undermined statutory protections by unduly restricting the time period in which victims can challenge discriminatory compensation. The drafting lineage is visible: Congress learned in 1991 that stating the override in the text prevents later argument about what the amendment was for, and applied the lesson.

The Civil Rights Restoration Act of 1987 likewise recites that certain aspects of recent decisions had unduly narrowed or cast doubt upon the broad application of four statutes, and states its purpose as restoring the prior institution-wide interpretation.

For a researcher, these sections are a shortcut. Where an amending statute states which decision it answers, the amendment history can be reconstructed from the enacted text alone, without reconstructing it from committee reports, and the reliability of that reconstruction is far higher.

The oddities in the 1991 text

Three features of Public Law 102-166 surprise readers who go to the text, and each one repays a moment of attention.

The first is the provision addressing the parties in the litigation that produced Wards Cove. The statute contains language that has the effect of excluding certain claims arising from that specific dispute from the new rules. Congress reversing a construction while carving out the case that produced it is unusual, and it reflects the ordinary reality that a negotiated statute accumulates provisions individual members require. A reader who finds it in the text and cannot find it in any summary is not misreading; the summaries omit it.

The second is the interpretive-memorandum provision limiting the legislative history usable in construing the business necessity language. Congress ordinarily produces legislative history and leaves courts to weigh it. Here it did the opposite, specifying that a designated statement is the exclusive legislative history for that phrase. The device exists because the parties agreed on words without agreeing on meaning, and it is one of the clearest artifacts of negotiation visible in any statute in this series.

The third is the placement of the damages provision at 42 U.S.C. section 1981a rather than within the Title VII chapter. There is a drafting logic to it, since the section serves claims under more than one statute, but the practical effect is that the most searched-for provision of the 1991 act sits away from the sections a reader would look at first, and researchers regularly conclude that the caps are not in the code because they looked in the wrong chapter.

Where each amendment shows up in a modern case file

A practical way to hold this history is to notice where each amendment appears in the documents an employment case actually generates.

The charge form and the deferral to a state agency reflect the 1964 design as modified in 1972, including the filing periods that depend on whether a state or local agency with jurisdiction exists.

The determination that the employer is covered turns on the fifteen-employee threshold from 1972 and, if the employer is a school district or a municipality, on the coverage extension from the same statute.

If the claim is brought by the commission rather than by the individual, that is the 1972 litigation authority.

If the complaint pleads pregnancy, the operative definition is the 1978 amendment.

If the complaint pleads a selection device rather than a decision about the individual, the burden allocation and the particular-practice requirement are the 1991 amendments, and the phrase job related for the position in question and consistent with business necessity is quoted directly from that statute.

If the complaint pleads that a protected characteristic was a motivating factor, that phrase is the 1991 amendment, and the employer’s answer asserting it would have taken the same action is the companion provision.

If the demand includes compensatory or punitive damages and a jury, that is 1991, and the applicable ceiling depends on the employer’s headcount under the tiers set that year.

If the claim is about pay set years earlier, its survival depends on the 2009 accrual rule.

And if the employer is a federal contractor and a compliance review is running alongside the charge, that is not this statute at all but the executive order issued in September 1965, which is a separate authority with a separate enforcement office.

Nine documents, six statutes, one act. That is the operative reality this article exists to describe, and it is why the answer to whether the Civil Rights Act of 1964 is still in effect is yes, in a form its drafters would recognize in outline and would have to be briefed on in detail.

The override cycle as a general phenomenon

The pattern this article names is not unique to civil rights law, but the frequency here is unusual, and the conditions that produce it can be stated.

An override requires four things. A judicial construction that a legislative coalition disagrees with. A harm from that construction that can be described to a non-lawyer in a sentence, since a technical burden-shifting rule is a harder sell than a woman who learned about years of underpayment too late to sue. An organized constituency with the capacity to keep the issue on the calendar across a Congress or two. And a vehicle, meaning either a standalone bill with enough support or a larger measure to attach to.

Measured against those four, the pattern in this statute makes sense. The pregnancy override had all four and took two years. The Ledbetter override had all four in unusually strong form, since the harm was a single sympathetic fact pattern, and it took eighteen months. The 1991 override had the first three and struggled on the fourth, requiring a veto, a failed override attempt and a renegotiation across two Congresses. The Title VI effects override has the first and arguably the third and has never had the second in a legible form, since the harm from losing a private right of action to enforce a regulation is real but difficult to state in a sentence, and it has not passed.

Two further observations follow. Overrides tend to be partial, because the coalition that assembles to reverse a decision includes members who want less than a full reversal, and the price of their votes is written into the text. Section 107 of the 1991 act is the clearest instance: liability was restored on more favorable terms than the decision allowed and the remedy was cut back, and both halves reflect real votes. And overrides tend to leave the reversed decision’s vocabulary in the statute, since the fastest way to reverse a holding is to write against it, which is why the amended text carries phrases like particular employment practice and not capable of separation for analysis that make no sense without knowing what they were written against.

How to date any claim about this statute

Six amendment dates and one construction date give a reader a complete diagnostic, and applying it resolves most disputes about what the act requires.

Ask first which title the claim concerns, because Title II is unchanged from 1964, Title VI changed once in 1988, and Title VII changed in 1972, 1978, 1991 and 2009.

Ask second whether the claim concerns coverage, proof, remedy or timing. Coverage questions are answered by 1964 and 1972 for employment and by 1964 and 1988 for federal funding. Proof questions are answered by 1991, with the caveat that the specific-practice requirement survived. Remedy questions are answered by 1991, and the answer differs depending on whether the claim is intentional discrimination or disparate impact. Timing questions are answered by 1991 for seniority systems and by 2009 for compensation.

Ask third whether the source is describing enacted text or judicial construction, since only the first appears in the Statutes at Large and only the second can be revisited by a court without legislation.

Ask fourth what date the source was written, because a treatise, article or opinion accurately describing the law in 1990 will describe a statute with no damages, no jury trial, an employer production burden in impact cases, no coverage abroad, and a pregnancy definition that had been in place for twelve years. Nothing in that description is an error; it is a description of a different text.

A reader who runs those four questions will not confuse the 1964 act with the 1991 act, will not assume punitive damages were always available, will not assume the caps are indexed, and will not cite a superseded standard as current law. Those four are, in order, the most common mistakes in writing about this statute.

For a reader building a working file on the amendment history, with each amending statute, the decision it answered and the code sections it touched kept in one place, you can keep your statute notes, citations and amendment chronologies together free on VaultBook.

The pairs that get confused

Four pairs account for most of the errors in writing about this history, and each has a clean distinguishing test.

The Civil Rights Act of 1964 and the Civil Rights Act of 1991. The first is a freestanding statute of eleven titles covering public accommodations, public facilities, education, federally assisted programs, employment and more. The second is principally an amending statute that changes the employment title of the first and the Reconstruction-era contracts statute, plus two titles of its own. A claim brought by a worker is brought under Title VII of the 1964 act as amended, not under the 1991 act, even when every operative rule in play was written in 1991. The test: if the sentence describes a prohibition, it belongs to 1964; if it describes a burden, a remedy or a cap, it probably belongs to 1991.

The Civil Rights Act of 1991 and the vetoed bill of 1990. They are different measures with different texts, and only one is law. The 1990 bill would have restored the business necessity burden and added damages, was vetoed on October 22, 1990, and failed an override attempt by one vote. Writing that Congress passed a civil rights act in 1990 is wrong; Congress passed a bill and it did not become law.

Title VII and the Reconstruction-era contracts statute at 42 U.S.C. section 1981. They overlap for race claims and differ in every practical respect: section 1981 has no administrative exhaustion requirement, a longer limitations period and no damages cap, while Title VII covers more protected characteristics, applies to employers of fifteen or more, and requires a charge before suit. Both were amended in 1991, and a decision construing one is not authority for the other. This is the single most common misattribution in secondary writing about the 1989 term, because Patterson v. McLean Credit Union is a section 1981 case that appears in every list of Title VII decisions Congress overrode.

The Civil Rights Restoration Act of 1987 and the year it was enacted. Its short title carries 1987 and it became law on March 22, 1988 over a veto. A citation giving either year is defensible if it says which it means, and a citation that mixes them without noticing is a signal that the writer took the date from a summary rather than from the statute.

The state-law layer, and why the federal amendments do not tell the whole story

A reader tracking only the federal amendments will misjudge what protection actually exists in a given workplace, because state law has moved on its own track and is frequently broader.

Many states cover employers below the federal fifteen-employee threshold, some down to a single employee. Many state statutes carry no damages caps, or higher ones, which is why plaintiffs’ counsel often plead state claims alongside federal ones and why the practical value of a claim varies by jurisdiction far more than the federal text suggests. Several states added protected characteristics decades before the federal construction of because of sex reached the same result in 2020, and several cover characteristics the federal statute does not reach at all.

Two consequences follow for the amendment history. The federal caps set in 1991 and never indexed are less binding in states with independent remedies, which reduces the pressure to revisit them, and that is a plausible partial explanation for legislative inaction on indexing. And the federal deferral arrangements, under which a charge filed in a state with its own fair employment agency runs on different deadlines, mean that the procedural rules a worker faces depend on where the workplace sits, a structure created in 1964, modified in 1972, and unchanged since.

The federal statute is a floor. The amendment history described in this article is the history of that floor rising, and the ceiling in any particular workplace is set by whichever body of law reaches further.

What a reader should be able to state after this

Six statutes, six dates, and one distinction.

The Equal Employment Opportunity Act of 1972 lowered the threshold to fifteen, added public employers and educational institutions, gave the commission authority to sue, and created the federal-employee process. The Pregnancy Discrimination Act of 1978 wrote pregnancy into the definition of sex discrimination, reversing a 1976 decision. The Civil Rights Restoration Act of 1987, enacted in 1988 over a veto, defined program or activity institution-wide across four statutes, reversing a 1984 Title IX decision. The Civil Rights Act of 1991 reversed or modified six decisions and created capped damages and jury trials. The Lilly Ledbetter Fair Pay Act of 2009 reset the pay-claim clock, reversing a 2007 decision. And the vetoed bill of 1990 is not among them, because it never became law.

The distinction is between an amendment, which changes the words and appears in the Statutes at Large, and a construction, which determines what the existing words mean and does not. A reader who holds the six statutes and that distinction can answer nearly any question about the difference between the act Congress passed and the law that operates.

Three things the ledger settles that arguments usually do not

Public argument about this statute recurs on three questions, and the amendment record answers all three without needing any interpretive judgment.

Whether the act is still law. It is, and the form it takes is set out above. Anyone asserting that it was repealed, expired, or replaced is wrong about a fact that the United States Code answers directly, and the answer does not depend on how the assertion is meant.

Whether the Supreme Court gutted it. The Court has never invalidated an operative provision. It narrowed six constructions that Congress then reversed, modified one that Congress adjusted, and produced one narrowing under Title VI that stands. Describing that record as gutting overstates it; describing it as leaving the statute untouched understates it. The precise version is that judicial construction has moved the statute in both directions and that the legislature has had the last word on most of the movement.

Whether the 1991 act was a win or a loss for its supporters. Both descriptions are used and neither survives the text. Supporters obtained a codified impact standard with the burden on the employer, damages and juries for intentional discrimination, extraterritorial coverage and four other reversals. They accepted caps that have not moved since, the exclusion of impact claims from damages, and a mixed-motive rule that limits remedies. A settlement with substantial gains and permanent ceilings is not well described by either word, and reaching for one of them is usually a sign that the speaker has a prior conclusion.

What the amendment history predicts about the statute’s future shape

Prediction is not the business of this series, but the pattern established across six enactments supports one structural observation that is not a forecast about any particular Congress.

Where a judicial construction produces a harm that can be stated in a sentence, affects an identifiable group with organizational capacity, and can be reversed by narrow drafting, this statute has been amended. Where any of those conditions fails, the construction has stood. The Title VI effects holding of 2001 fails the first and third conditions: the harm is the loss of a private vehicle to enforce a regulation, which does not compress into a sentence a non-lawyer will act on, and reversing it requires drafting a private right of action with its own standards and remedies rather than striking a phrase.

The damages caps illustrate the same conditions operating on an omission rather than on a decision. Nobody enacted the erosion of the caps; it happens because nominal ceilings do not adjust. There is no decision to point at, no dated harm, and no single case to name, which is exactly the profile of a problem that persists through legislative inaction.

The observation generalizes to any statute in this series. Legislative correction is not a function of how serious a problem is. It is a function of how legible the problem is, how narrowly the fix can be drafted, and whether a constituency exists to hold the calendar. That is a description of the mechanism, not a judgment about the outcomes it produces.

Does the operative statute still resemble what Congress passed in 1964?

In structure yes, in detail no. The eleven titles, the prohibitions and the enforcement architecture are recognizably the 1964 design. The coverage threshold, the covered employers, the enforcement authority, the proof standards, the remedies and the accrual rules have all changed by statute since.

How much of the operative employment law was written after 1964?

Most of the machinery. The prohibitions are 1964, but the coverage threshold, the covered employers, the agency’s litigation authority, the disparate impact burden, the motivating-factor rule, the damages, the jury right, the extraterritorial reach and the pay-claim accrual rule were all enacted between 1972 and 2009.

Stating the proportion that way is not a criticism of the original statute. A first enactment establishes prohibitions and an enforcement design, and the design is where the compromises land, because a prohibition is easy to vote for and an enforcement mechanism is where the cost falls. Congress in 1964 wrote prohibitions strong enough that no later Congress has needed to strengthen them and a mechanism weak enough that every later Congress that revisited the statute strengthened some part of it. That asymmetry is the most transferable observation in this article, and it appears again and again across the statutes in this series.

Reading the enrolled texts rather than the summaries

A short methodological note belongs here, because the errors this article corrects are almost all downstream of reading summaries.

The enrolled text of each amending statute is short. The Pregnancy Discrimination Act is a single substantive sentence plus a proviso. The Ledbetter Act is roughly two pages. The 1972 statute and the 1991 statute are longer but still readable in an afternoon, and both are organized by section with descriptive headings. Reading them directly produces three things a summary cannot.

It shows what changed and what did not. A summary says the 1991 act strengthened Title VII. The text shows that section 105 allocated a burden, section 102 created and capped a remedy, section 107 traded liability for remedy, and section 106 prohibited a practice that had nothing to do with any of the decisions being reversed.

It shows the negotiation. The interpretive-memorandum provision, the carve-out for particular parties, and the placement of the damages section elsewhere in the code are all visible in the text and invisible in every summary.

It shows the drafting technique. Amending a definition reaches every provision using the term; amending an operative section reaches only that section; enacting a standalone statute requires building enforcement machinery from scratch. Congress used all three approaches across these six statutes, and comparing them teaches more about statutory design than any general treatment.

Where does a reader find the text of each amending statute?

Each amending act has a public law number and a Statutes at Large citation, and the codified result carries credits identifying every public law that touched the section. Moving from the code section to the credits to the enrolled text is the reliable route, and it catches provisions the code’s arrangement separates from their source.

One further habit is worth adopting alongside the reading. Whenever a source describes an obligation under this act, check whether it names a year, and if it does not, supply one before relying on the description. A sentence stating that Title VII covers employers with a certain number of employees, that a plaintiff may or may not recover damages, that pregnancy is or is not within the definition of sex, or that a pay claim accrues at a particular moment is a sentence about a period. Six of those propositions have had two correct answers at different times, and the interval between the answers runs from two years to forty-five. Adding the year costs nothing and eliminates the largest category of error in writing about this statute, which is not misunderstanding the law but describing the wrong version of it accurately.

A related caution applies to secondary sources written between 1989 and 1991, and between 2007 and 2009. Both windows sit between a narrowing decision and its legislative reversal, and material produced in them describes a statute that operated for a short, defined period. Such sources are not unreliable; they are dated, and dating them is the reader’s job rather than theirs.

A closing note on what this history is evidence of

The six amendments are usually presented as a story about civil rights politics, and they are that. They are also evidence about something more general, which is why this article belongs in a series about how American statutes work rather than only in a cluster about one act.

A statute is not finished when it is signed. It is handed to agencies that write rules, to courts that construe text, and to regulated parties who respond to incentives, and the legislature that wrote it retains the power to correct any of them and exercises that power unevenly. The 1964 act has been corrected six times in response to judicial construction, once in response to an enforcement failure the legislature could observe directly, and not at all in response to the one narrowing that lacked a legible constituency. Each of those outcomes is explained by the same three variables, and none of them is explained by the strength of anyone’s commitment to the statute’s purposes.

That is the claim this article defends, and the ledger above is the evidence for it.

The operative law, stated as it stands

The closing statement this framework demands is a description of the law after all of it, and it can be given in one pass.

The public accommodations title reaches inns and hotels, restaurants and other food service establishments, theaters and other places of exhibition or entertainment, and establishments physically located within covered establishments, with a private club exemption and a small owner-occupied lodging exemption. It is enforced by injunction with attorney’s fees and carries no damages. It has not been amended.

The federally assisted programs title prohibits discrimination on the ground of race, color or national origin in any program or activity receiving federal financial assistance, with program or activity defined institution-wide since 1988. It is enforced administratively through agency regulations and fund termination, and by private suit for intentional discrimination only, since a 2001 decision eliminated private enforcement of the effects regulations and Congress has not reversed it.

The employment title reaches employers with fifteen or more employees, including state and local governments and educational institutions, with a separate process for federal employees and, since 1995, for legislative-branch employees. It prohibits discrimination because of race, color, religion, sex or national origin, with sex including pregnancy since 1978 and encompassing sexual orientation and gender identity as a matter of construction since 2020. Liability attaches where a protected characteristic was a motivating factor, subject to a defense that limits remedies where the employer would have acted the same way. It reaches practices that cause a disparate impact unless the employer demonstrates job relatedness and business necessity, with the plaintiff required to identify a particular practice except where the elements of a process cannot be separated. The enforcement commission investigates, conciliates and may sue. A prevailing plaintiff may obtain injunctive relief, reinstatement, back pay, attorney’s fees and expert fees, and, for intentional discrimination, compensatory and punitive damages capped by employer size with a jury trial right, with punitive damages unavailable against government employers. Compensation claims accrue with each paycheck issued under a discriminatory decision. Coverage extends to American citizens employed abroad by American-controlled employers subject to a foreign-law exception.

Not one of those three paragraphs could be written from the 1964 text alone, and the second and third could not be written from the 1964 text plus the case law either. That is the point of this article. The statute in force is the enacted text plus every amendment plus every surviving construction, and a reader who has the ledger has the map. For orientation across the whole act, the complete guide to the Civil Rights Act of 1964 is the hub this cluster reports to.

Frequently Asked Questions

Q: Is the Civil Rights Act of 1964 still in effect?

Yes, and in substantially amended form. Public Law 88-352 has never been repealed and no operative provision has been held unconstitutional. Its employment title has been amended four times: in 1972, which lowered the coverage threshold to fifteen employees, added state and local governments and educational institutions, and gave the enforcement commission authority to sue; in 1978, which brought pregnancy within the definition of sex discrimination; in 1991, which allocated the disparate impact burden to the employer, added a motivating-factor rule, and created capped damages with a jury trial right; and in 2009, which changed when a pay claim accrues. Title VI was amended in 1988 to define program or activity institution-wide. Title II has never been amended.

Q: What did the Civil Rights Act of 1991 change?

It responded to six Supreme Court decisions and filled a gap that had existed since 1964. It codified disparate impact analysis and placed on the employer the burden of demonstrating job relatedness and business necessity once a plaintiff identifies a particular practice. It provided that liability attaches where a protected characteristic was a motivating factor, while allowing an employer that proves it would have acted the same way to avoid damages and reinstatement. It created compensatory and punitive damages for intentional discrimination with a jury trial right and caps by employer size. It extended the Reconstruction-era contracts statute to post-formation conduct, extended Title VII to Americans employed abroad by American-controlled employers, reset the seniority-system clock, limited collateral attacks on consent decrees, and barred score adjustment on employment tests.

Q: What did the Equal Employment Opportunity Act of 1972 change about Title VII?

Four things. It lowered the employer threshold from twenty-five employees to fifteen. It removed the exclusions for state and local governments and for educational institutions, bringing police and fire departments, school districts and public universities within the statute for the first time. It gave the Equal Employment Opportunity Commission authority to bring civil actions in its own name when conciliation failed, transferring enforcement cost from the individual complainant to the government in cases the agency selected. And it created a separate administrative process for federal employees with eventual access to federal court. The commission had sought cease-and-desist authority instead; the court-enforcement model was the compromise that passed.

Q: Why was the Civil Rights Act of 1990 vetoed?

President Bush vetoed the bill on October 22, 1990. The veto message argued that its business necessity language and its damages provisions would pressure employers to adopt hiring and promotion practices keyed to numerical outcomes in order to limit litigation exposure, using the word quotas for that pressure, while stating support for stronger remedies against intentional discrimination. The sponsors, principally Senator Kennedy and Representative Hawkins, responded that the bill restored the burden allocation that had governed from 1971 to 1989 without producing that effect, and that requiring justification for a practice that excludes disproportionately is not a requirement to hire by the numbers. The Senate override failed by one vote on October 24, 1990.

Q: How does the Pregnancy Discrimination Act relate to Title VII?

It is an amendment to Title VII’s definitions section rather than a separate statute, which is why it reaches every provision in which the defined term appears. It was enacted in 1978 as Public Law 95-555 in response to General Electric Co. v. Gilbert, 429 U.S. 125 (1976), in which the Supreme Court held that excluding pregnancy from a disability plan was not discrimination because of sex. The amendment provides that because of sex includes because of pregnancy, childbirth or related medical conditions, and that women affected by those conditions must be treated the same for all employment-related purposes as other persons similar in their ability or inability to work. That comparison is the analytical hinge in subsequent litigation.

Q: How did the Lilly Ledbetter Act amend the Civil Rights Act of 1964?

It changed when a pay discrimination claim accrues, and nothing else. Public Law 111-2, signed on January 29, 2009, provides that an unlawful practice occurs when a discriminatory compensation decision is adopted, when an individual becomes subject to it, and each time compensation is paid pursuant to it. It reversed Ledbetter v. Goodyear Tire and Rubber Co., 550 U.S. 618 (2007), which had held that the filing clock runs from the compensation decision itself. It applies parallel amendments to the age, disability and rehabilitation statutes and covers claims pending on or after May 28, 2007. It did not create an equal pay guarantee, change what counts as pay discrimination, or remove the two-year limit on recoverable back pay.

Q: Have any parts of the Civil Rights Act of 1964 been repealed?

No title has been repealed and no operative provision has been struck down. What has changed is the content of several provisions through amendment, and the practical reach of one through judicial construction that Congress has not reversed. The employment title’s threshold, coverage, proof standards, remedies and accrual rules all differ from the 1964 text. Title VI’s definition of program or activity was narrowed by a 1984 decision and restored by statute in 1988, and its effects regulations lost private enforceability in Alexander v. Sandoval in 2001, which remains the operative rule. Title II stands exactly as enacted, including its private club and small-lodging exemptions.

Q: Which later statutes expanded the Civil Rights Act of 1964?

Four amended it directly: the Equal Employment Opportunity Act of 1972, the Pregnancy Discrimination Act of 1978, the Civil Rights Restoration Act of 1987 enacted in 1988, and the Civil Rights Act of 1991, with the Lilly Ledbetter Fair Pay Act of 2009 adding the compensation accrual rule. Several others changed how it operates without amending it: the Congressional Accountability Act of 1995 applied Title VII to legislative-branch employees, the Civil Service Reform Act of 1978 reorganized federal-sector administration, and the age, disability and genetic information statutes borrow its charge machinery while carrying their own causation standards. The Fair Housing Act of 1968 and the Voting Rights Act of 1965 are separate statutes, not expansions of this one.

Q: Are the 1991 damages caps adjusted for inflation?

No. The statute set four tiers by employer size, and none carries an indexing provision or has been raised since November 1991, so the real value of the ceiling has declined every year since. The combined cap on compensatory and punitive damages per complaining party runs from fifty thousand dollars for employers with more than fourteen and fewer than one hundred and one employees up to three hundred thousand dollars for employers with more than five hundred. Back pay falls outside the cap because it is an equitable make-whole remedy rather than damages. Punitive damages are unavailable against a government, government agency or political subdivision, and damages of any kind are unavailable in disparate impact cases.

Q: Did the 1991 act simply strengthen the 1964 act?

Not simply. It restored the employer’s burden in disparate impact cases, reversed five other constructions, and created damages and jury trials that had never existed. It also capped those damages, withheld them from disparate impact claims entirely, retained the specific-practice requirement from the decision it was written against, and converted the mixed-motive rule into a compromise in which liability attaches on a motivating-factor showing while an employer’s same-decision proof cuts off damages, reinstatement and back pay. Every one of those limits was the price of the votes that passed it. Describing the statute as a restoration or as a rollback both miss; it was a negotiated settlement with real gains and real ceilings.

Q: Does the Civil Rights Act of 1991 apply to cases filed before it passed?

No, as to its damages and jury trial provisions. The statute’s effective-date language was ambiguous because the drafters could not agree, and the Supreme Court resolved the question in Landgraf v. USI Film Products, 511 U.S. 244 (1994), holding that those provisions do not apply to conduct occurring before enactment, and setting out the general framework for statutory retroactivity questions. Rivers v. Roadway Express, decided the same day, applied the same reasoning to the amendment of the contracts statute. The practical result is that reported decisions from the mid-1990s applying superseded standards are usually correct retroactivity outcomes rather than errors. Congress wrote an explicit application date into the 2009 statute to avoid a repeat.

Q: Why has Title II never been amended?

Because the Supreme Court upheld it immediately and then had few occasions to construe it narrowly, so there was nothing for Congress to reverse. Heart of Atlanta Motel and Katzenbach v. McClung closed the constitutional question in December 1964, a 1968 decision made attorney’s fee awards routine for prevailing plaintiffs, and a 1969 decision closed the private club evasion for places of entertainment. The title also generated little litigation, since it offers injunctive relief and fees but no damages, so a person refused service has no financial claim to pursue. Proposals to widen its covered categories or add damages have been introduced across the decades without passing, leaving federal public accommodations coverage narrower than the law of many states.

Q: What is the difference between an amendment and a court decision changing the statute?

An amendment changes the words. It passes both chambers, goes to the President, receives a public law number, and appears in the Statutes at Large, and a reader can compare the old text with the new one. A judicial construction determines what the enacted words mean without changing them. Three consequences follow. A construction can be revisited by courts in a way an enacted text cannot. A construction can be reversed by ordinary legislation, which is what four of the entries in this article’s ledger did. And research that traces only the amending statutes will miss constructions entirely, which is why the amendment history and the litigation history have to be read together.

Q: Which Supreme Court decisions has Congress overridden by amending this act?

Six, across four statutes. General Electric Co. v. Gilbert (1976) on pregnancy, overridden in 1978. Grove City College v. Bell (1984), a Title IX decision construing shared language, overridden in 1988 across four civil rights statutes. Wards Cove Packing Co. v. Atonio, Patterson v. McLean Credit Union and Lorance v. AT&T Technologies, all from 1989, plus EEOC v. Arabian American Oil Co. from March 1991, overridden by the Civil Rights Act of 1991, which also modified Price Waterhouse v. Hopkins and limited the collateral attacks permitted by Martin v. Wilks. And Ledbetter v. Goodyear (2007), overridden in 2009. Alexander v. Sandoval (2001) is the significant narrowing Congress has not answered.

Q: What enforcement power does the EEOC have that it did not have in 1964?

The authority to sue. As enacted, Title VII gave the commission power to investigate charges and attempt conciliation and nothing more, so a failed conciliation produced a notice to the complainant rather than a remedy, and the worker then needed private counsel to reach court. The Equal Employment Opportunity Act of 1972 gave the commission authority to bring civil actions in its own name, which created a public enforcement track capable of pursuing systemic pattern-or-practice cases that individual plaintiffs could not finance. Congress considered giving the commission cease-and-desist authority to order relief administratively, as federal labor law does, and chose court enforcement instead. That choice shaped every subsequent decade of the statute’s operation.

Q: How should a researcher trace an amendment through to the operative text?

Start from the code section, read its credits to identify each public law that touched it, then read the amending act’s own text in the Statutes at Large rather than relying on the codified version, because one amending statute often changes several sections at once and the code’s arrangement can obscure the connection. Watch for provisions that were placed outside the obvious chapter: the 1991 damages provision sits at 42 U.S.C. section 1981a, next to the Reconstruction-era contracts statute rather than with the rest of the employment title, and a search confined to the 2000e sections will not find it. Then check whether any construction of the amended language has narrowed it since enactment.