The One-Paragraph Answer

A mailed ballot is a paper ballot that travels through the postal system instead of being handed to a voter at a polling place, and its security rests on a chain of layered checks rather than on any single safeguard. The voter requests or automatically receives a ballot, marks it in private, seals it in a signed envelope, and returns it by mail or to a secure drop box. Election workers then verify the signature against the registration record, open only the accepted envelopes, and count the ballots on the same tabulation equipment used for in-person votes, with the results checked afterward by audits. The accumulated research finds that mail voting raises turnout modestly, on the order of a few percentage points, and does not systematically favor either major party. Documented fraud exists but is rare, and the fiercest disputes about the practice are better understood as arguments about how much trust a voting system should require than as disputes about what the record shows.

The thesis is that vote-by-mail is a documented administrative process whose integrity comes from redundancy, meaning that no one step carries the whole burden of proof, and that the evidence consistently shows modest turnout gains without a partisan tilt, which makes the loudest fights about the practice arguments about institutional trust rather than about fraud the record can demonstrate. That claim is worth stating plainly because the public conversation usually collapses two different questions, how the machinery works and whether the machinery can be trusted, into a single slogan. The machinery is knowable. Election offices publish their procedures, researchers measure the outcomes, and the failure modes have names and dates. Trust is a separate question, and it cannot be settled by a procedure manual. The machinery is described in full below, with what the evidence shows about fraud and participation, the strongest version of the case against mail voting as well as the case for it, and then the reason the evidence has not settled the argument.

What follows is written for a reader who has heard mail voting described as both perfectly safe and fundamentally corrupt and wants the administrative reality underneath the slogans. The payoff is practical: by the end, that reader will be able to trace a mailed ballot from the election office to the voter and back, name the safeguard applied at each step and the known limit of that safeguard, and evaluate any future claim about fraud or turnout against the documented record rather than against a partisan script. The first sections walk the ballot’s journey in order. The middle sections cover the federal rules and the history. The later sections report the evidence on fraud and turnout, state the strongest critique and the strongest defense, and close with the honest complication, which is that a voting system can be both measurably secure and widely distrusted, and no procedure fixes that gap by itself.

The Three Designs of Mail Voting

The phrase vote-by-mail describes three different systems, and most confusion about the practice starts with treating them as one. In the first design, every registered voter automatically receives a ballot in the mail for every election, with no request needed. Eight states use this universal model: Oregon, Washington, Colorado, Utah, Hawaii, California, Vermont, and Nevada. In the second design, any registered voter may request a mailed ballot without giving a reason, and the ballot goes only to those who ask. This no-excuse absentee model covers most of the remaining states. In the third design, a voter must state a qualifying reason, such as illness, disability, age, or absence from the county on Election Day, before the office will mail a ballot. These excuse-required states are the smallest group. The three designs share the same return and counting machinery described below, but they differ in one decision that shapes everything else: whether the ballot goes out automatically or only on request.

The distinction matters because the risks and the administrative burdens attach to different stages in each design. In universal systems, the election office must keep its address files current, since ballots mailed to stale addresses are the system’s largest source of undeliverable mail. The safeguard is list maintenance: regular comparison against postal change-of-address records, motor vehicle files, and death records, plus the return of undeliverable ballots as a signal that a registration needs attention. In request-based systems, the address problem is smaller because the voter supplies a current mailing address with each request, but the office must process millions of individual requests on a deadline, which concentrates the workload into a few frantic weeks. In excuse-required systems, the office must also adjudicate the stated reasons, which adds a layer of judgment and paperwork. None of the three designs is the default against which the others are measured. Each is a complete system with its own failure points, and the evidence about fraud and turnout has to be read with the design in mind, because a finding about one system does not automatically transfer to another.

The terminology around these systems is worth pinning down, because the words carry different legal meanings in different states. Absentee voting originally meant voting away from one’s assigned polling place, and the term survives in state statutes even where no excuse is required. All-mail voting and vote-by-mail usually describe the universal design. Early voting describes in-person voting before Election Day and is a separate practice, though it is often confused with mail voting because both shift votes away from a single Tuesday. Military and overseas voters vote under a federal framework, the Uniformed and Overseas Citizens Absentee Voting Act of 1986, that exists alongside every state system. When vote-by-mail is the term in use, it means any system in which the voter receives a paper ballot through the mail, marks it outside a polling place, and returns it for counting. When a claim depends on which design is in force, the design is named each time.

Who Receives a Mailed Ballot

Eligibility to receive a mailed ballot starts with voter registration, and the rules for who is on the rolls are the foundation the whole system stands on. Every state maintains a voter registration list, and federal law sets floors for how those lists are built and maintained. The National Voter Registration Act of 1993 requires states to offer registration at motor vehicle offices, by mail, and at public assistance agencies, while limiting how aggressively states may remove names from the rolls. That statute’s registration channels and its purge protections shape who appears on the list from which mailed ballots are drawn, and its full mechanics are covered in the companion guide to the registration law. The Help America Vote Act of 2002 added a second federal floor by requiring statewide computerized registration lists and setting identification rules for first-time voters who registered by mail. Those federal requirements apply inside every mail voting system, whether the state mails ballots to everyone or only to those who ask.

In universal systems, the mailing list is the registration list itself, filtered for active status. Voters marked inactive, usually because mail sent to them bounced or because they have not voted or responded to confirmation notices across two federal election cycles, do not automatically receive ballots in most universal states, though the details vary. In request-based systems, the voter must be registered and must then file a request, which typically asks for name, address, date of birth, and a signature, and in some states an identification number such as a driver’s license number or the last four digits of a Social Security number. The office checks the request against the registration record before issuing a ballot. This check is the first verification gate: a ballot is never supposed to leave the office for a name that does not match a registered voter, or for an address the office cannot tie to that voter. Where the request is defective, the office is supposed to notify the voter and offer a correction rather than simply issuing or refusing the ballot.

Two special populations vote by mail under rules that differ from the domestic ones. Members of the military, their families, and civilians living abroad vote under the Uniformed and Overseas Citizens Absentee Voting Act, which requires states to send them ballots and, since the Military and Overseas Voter Empowerment Act of 2009, to transmit requested ballots at least 45 days before a federal election. Voters with disabilities are entitled under federal law to accessible voting, and several states have extended electronic ballot delivery, though not electronic return, to voters who cannot mark a paper ballot independently. The details of those accommodations sit outside the standard mail voting pipeline, but the counting machinery at the end is the same. For the ordinary domestic voter, the pipeline begins with a simple administrative question, is this person registered and entitled to a ballot for this election, and the rest of the system is built to make sure the answer given at the start is the answer honored at the count.

Request and Issuance: How the Ballot Leaves the Office

Once eligibility is established, the election office assembles the ballot packet, and the care taken at this stage determines how much trust the later stages can carry. A standard packet contains the ballot itself, printed for the voter’s specific precinct and ballot style so that the voter receives exactly the contests they are entitled to vote in; a secrecy envelope or sleeve, used in many states to separate the voter’s identity from the marked ballot; and a return envelope, pre-addressed to the election office, that carries the voter’s identifying information and signature line. Many offices now print an Intelligent Mail barcode on the return envelope, which lets both the office and the voter track the ballot through the postal system the way a parcel is tracked. The ballot styles matter more than they appear to: in a single county there may be dozens of combinations of congressional, legislative, and local districts, and issuing the wrong style to a voter is one of the most consequential clerical errors an office can make, because it can only be corrected by reissuing the ballot before the deadline.

The office logs every ballot it issues, and that log becomes the backbone of the system’s accounting. Each issued ballot is recorded against the voter’s registration record, which means the office knows exactly how many ballots went out, to whom, and in which ballot style. When a voter later returns a ballot, the office checks the return against the issuance log, and a second ballot arriving under the same voter’s name is flagged rather than counted twice. If a voter reports that a ballot never arrived or was spoiled, the office can cancel the original in the log and issue a replacement, and the canceled ballot becomes uncountable even if it later surfaces. This accounting is the quiet workhorse of mail voting security. It does not depend on trusting the voter or the postal system. It depends on a database that records every ballot as a unique, trackable object, and it is the reason that the nightmare scenario of photocopied ballots flooding the count fails at the first step, because a ballot the office never issued has no record to match and is rejected on receipt.

Timing rules govern when ballots may be mailed, and the calendar is tighter than most voters realize. Federal law requires ballots for overseas and military voters to go out at least 45 days before a federal election. For domestic voters, states set their own mailing windows, commonly beginning three to four weeks before Election Day, which is why voters who request ballots late sometimes receive them with only days to spare. Offices are required to process requests received by statutory deadlines, which usually fall about a week before the election, and late requests are denied. The tight calendar is one reason election administrators urge voters to request early and return early, and it is also the source of a recurring administrative strain: a surge of last-minute requests can overwhelm small county offices, producing delays that then compress the voter’s window for returning the ballot. The system works best when the ballot spends most of its life in the voter’s hands rather than in transit, and every deadline in the pipeline is designed to protect that balance.

Marking, Sealing, and the Return Journey

The voter’s side of the process is the least supervised part of the system, and that is both the point and the problem. A mailed ballot is marked in private, at the kitchen table or the office desk, without a poll worker watching. The voter fills in the ovals or connects the arrows as instructed, places the ballot in the secrecy sleeve where the state provides one, seals the sleeve inside the return envelope, and signs the envelope where indicated. The signature is the voter’s credential. It is what the office will compare against the registration record, and it is the reason the return envelope, not the ballot, carries the voter’s name. The secrecy sleeve exists to preserve the secret ballot after the envelope is opened: workers verify the signature on the sealed envelope, and only after a ballot is accepted is the envelope opened and the sleeved ballot separated from the voter’s identity before counting.

The return trip offers the voter more choices than the phrase voting by mail suggests. The ballot may go back through the postal service, and most states require it to be postmarked by Election Day or received by a stated deadline, with the exact rule varying enough that voters who move between states get tripped up. It may be carried to a secure drop box, a locked receptacle bolted to concrete, typically under video surveillance, that is emptied on a schedule by election workers, often working in bipartisan pairs. It may be hand-delivered to the election office or, in some states, to a polling place on Election Day. What the voter may not do is hand the ballot to just anyone. Most states restrict who may return another voter’s ballot, usually to family members, household members, or designated agents, and a handful of states ban third-party collection outright. These collection rules, sometimes called ballot harvesting laws by their critics, are among the most contested details in mail voting administration, because they sit exactly on the boundary between helping voters participate and opening a channel for coercion or tampering.

The privacy of the home-marked ballot is the feature that distinguishes mail voting from every other method, and it is worth understanding precisely what is gained and lost. What is gained is convenience and time: the voter can research candidates and measures at leisure, consult sources, and vote without traveling or waiting. What is lost is the controlled environment of the polling place, where no one watches the voter mark the ballot and no one can stand over the voter’s shoulder. The system compensates with procedure rather than supervision. The signature check substitutes for the poll worker’s glance at an ID. The issuance log substitutes for the poll book. The secrecy sleeve substitutes for the privacy booth. Whether that substitution is adequate is the central dispute about mail voting, and it cannot be resolved by describing the procedure. It can only be addressed by asking what the procedure has actually prevented and what it has actually missed, which is the subject of the evidence sections below.

Deadlines, Postmarks, and Drop Boxes

Every mail voting system is a race against a clock, and the clock is set by statute, not by the post office. The controlling question is which moment counts: the moment the ballot is mailed or the moment it arrives. Most states require the ballot to be received by the election office by a stated deadline, commonly the close of polls on Election Day, and a ballot that arrives later is not counted no matter when it was mailed. A minority of states count ballots that are postmarked by Election Day and arrive within a grace period of several days. The distinction produces real consequences. In receipt-deadline states, a ballot mailed on time but delayed in transit is a lost vote through no fault of the voter. In postmark states, the voter bears less transit risk, but the office must hold the count open and verify postmarks, which slows the final tally. Litigation over which rule applies, and over whether a missing or illegible postmark disqualifies an otherwise valid ballot, has recurred in federal courts, and the answers differ by state and by election.

Drop boxes were designed to take the postal service out of the equation for voters who distrust the mail or who are voting close to the deadline. A drop box is a piece of election infrastructure, not a mailbox: it is owned or contracted by the election office, anchored in place, built to resist tampering, and emptied on a documented schedule. States regulate them with varying strictness. Some require continuous video surveillance and mandate that boxes be emptied by two workers of different parties. Some limit the number of boxes per county or require them to sit on government property. A few states have banned them. The security logic of the drop box is straightforward: it shortens the chain of custody by removing the postal system from it, and it gives the voter a receipt-like certainty that the ballot reached the office, since many tracking systems record the moment a box is emptied. The political logic is less straightforward, because the visibility of the boxes makes them a focal point for disputes about unsupervised ballot collection, even though the boxes themselves are among the most supervised links in the chain.

The deadline structure also creates the phenomenon of the late-decided voter, and the system’s answer to that voter is the in-person fallback. A voter who requested a mail ballot but never returned it can usually still vote in person, and the poll book check against the issuance log prevents the double vote. A voter who returned a mail ballot and then tries to vote in person will be stopped by the same check. A voter who spoiled a mail ballot can surrender it at the polling place and vote a fresh one. These fallback rules are essential to the system’s integrity, because they mean a lost or delayed mailed ballot does not disenfranchise the voter who can still reach a polling place. They also mean the issuance log is doing double duty, guarding both the mail pipeline and the in-person rolls against the same voter’s ballot being counted twice. The log is checked at the moment of return and again at the moment of any in-person voting attempt, and both checks have to clear before a vote is recorded.

From Mailbox to Election Office

When a mailed ballot arrives at the election office, it enters a processing pipeline that is industrial in scale and legalistic in detail. The first step is intake: workers, often temporary staff hired for the election season and trained on the specific procedures, collect the return envelopes from the mail room and the drop box runs, and log each envelope against the issuance record. The envelope is checked for completeness before anything else. Is it signed. Is the signature in the right place. Is the envelope sealed. In states that require a witness signature or a voter identification number on the envelope, those fields are checked too. An envelope that fails these threshold checks is set aside for the cure process or for rejection, and it is never opened. The unopened envelope is important: until the voter’s credentials are accepted, the ballot inside remains legally uncast, which means a rejected envelope is a rejected voter, not a rejected vote, and the distinction matters for how the voter is notified and what remedies are available.

The accepted envelopes move to signature verification, the most labor-intensive and most debated step in the pipeline. Workers compare the signature on the return envelope against the signature or signatures on file for that voter, usually the registration signature and sometimes the signatures from prior elections or motor vehicle records. The comparison is done by trained staff, and in larger jurisdictions the first pass is done by software that flags envelopes for human review. A signature that clearly matches is accepted. A signature that clearly does not match is rejected. The difficult cases are the ones in between, and offices handle them with layered review: a first worker’s rejection is typically reviewed by a second worker, and in many offices a bipartisan team makes the final call on challenged signatures. The standard the workers apply is not forensic certainty. It is whether the signature is consistent with the voter’s known signatures, allowing for the natural variation in how people sign their names over time and under different conditions. That standard is deliberately pragmatic, and its pragmatism is exactly what critics attack and defenders rely on.

Once the signature is accepted, the envelope is opened and the ballot is separated from the voter’s identity. In states with secrecy sleeves, the sleeved ballot goes into one stream and the signed envelope into another, so that by the time the ballot reaches the counting equipment, no one handling it can connect it to a voter. The envelopes themselves are retained, because they are the audit trail: if a count is ever challenged, the envelopes prove which voters’ ballots were accepted and allow investigators to re-verify the signatures. The ballots are then prepared for tabulation, which in most jurisdictions means they are scanned by the same optical-scan equipment that counts in-person paper ballots, or, where the office uses central-count equipment, batched and fed through high-speed scanners. Nothing about the counting step is specific to mail voting. The ballot, once separated from its envelope, is indistinguishable from any other paper ballot, and that indistinguishability is a security feature: it means the tabulation cannot treat mailed votes differently from votes cast in person.

Signature Verification: The Credential Check

How Does Signature Verification on a Mail Ballot Actually Work?

Election workers compare the envelope signature against the voter’s signatures on file, usually the registration signature plus records from earlier elections. Software does the first pass in large counties, and staff review every flag. A clear match is accepted, a clear mismatch goes to a second reviewer, and many offices require a bipartisan team for the final rejection.

Signature verification is the credential check of mail voting in the same way that showing identification is the credential check of in-person voting, and it inherits both the strengths and the weaknesses of any system that authenticates people by how they write their names. Its strength is that it ties every ballot to a specific registered voter without requiring the voter to appear anywhere or present anything beyond the envelope. Its weakness is that signatures are unstable. People’s signatures change with age, injury, haste, and the writing surface, and the reference signatures on file may be years old. Research on signature comparison has found substantial error rates even among trained examiners, which means the verification step inevitably produces two kinds of mistakes: accepting a signature that is not the voter’s, and rejecting a signature that is. The system is tuned, by training and by procedure, to prefer the second kind of mistake over the first, because a wrongly rejected ballot can be cured while a wrongly accepted one cannot be uncounted. That tuning is a policy choice, and it has consequences for who bears the cost of the system’s imperfection.

The uneven distribution of those costs is one of the best-documented problems in mail voting administration. Studies of rejected mail ballots have found that rejection rates run higher for young voters, for first-time mail voters, and for Black and Hispanic voters than for older white voters, a pattern researchers attribute to less stable signature histories, less familiarity with the cure process, and reference signatures that are older or less representative. The overall rejection rate for mail ballots in federal elections is typically around one percent, which sounds small until it is multiplied by tens of millions of mailed ballots, at which point it represents hundreds of thousands of voters whose ballots did not count. Defenders of the system point out that the cure process exists precisely to rescue those voters, and that states with robust cure procedures and voter education see lower effective rejection rates. Critics reply that a safeguard which systematically burdens some voters more than others is a defect in the safeguard, not a price of security. Both claims can be true at once, and the honest description of signature verification is that it is a workable credential check with a measurable and uneven error rate, not a perfect one.

The alternatives to signature verification illustrate why it persists despite its flaws. A few states require the voter to include a copy of an identification document or an identification number on the return envelope, which shifts the credential from a behavioral biometric to a knowledge check. Witness and notary requirements, still on the books in a small number of states, add a second human to the transaction but impose a cost in convenience that researchers have found depresses participation. Proposals to replace signatures with other credentials run into the same problem that created signature verification in the first place: any credential strong enough to satisfy the skeptics is burdensome enough to deter the voters the system is trying to serve. The signature endures because it is the credential every registered voter already possesses, it can be checked without the voter’s presence, and its failures are visible in the rejection data rather than hidden. A system whose errors can be counted is a system that can be improved, and the rejection statistics, uncomfortable as they are, are the reason the error rate is known at all.

The Cure Process: Rescuing the Rejected Ballot

A rejected mail ballot is not necessarily a lost vote, because more than two dozen states require election officials to notify the voter and offer a chance to fix the problem. The cure process, as administrators call it, typically works like this: when an envelope is rejected for a missing or mismatched signature, the office contacts the voter by mail, phone, email, or text message, explains the defect, and provides a cure affidavit, a sworn statement in which the voter attests that the ballot is theirs, sometimes accompanied by a copy of an identification document. The voter signs and returns the affidavit by a statutory deadline, which usually falls a few days after Election Day, and the ballot is then accepted and counted. The cure affidavit becomes part of the audit trail, attached to the original envelope, so that the rescue itself is documented. States without a cure requirement simply reject the ballot and notify the voter, if they notify the voter at all, which means the same signature defect disenfranchises a voter in one state and merely inconveniences a voter in another.

The cure process is where the administrative quality of a mail voting system shows most clearly, because it converts a binary pass-fail check into a conversation. Offices that run it well contact voters quickly, through multiple channels, in languages the voters read, and they staff phone lines to walk voters through the affidavit. Offices that run it poorly send a single letter that arrives after the cure deadline has passed. The difference shows up in the data: jurisdictions with aggressive cure programs report that a large share of initially rejected ballots are ultimately counted, while jurisdictions with weak programs report that rejection is effectively final. Civil rights litigators have made the cure process a recurring target, arguing that the absence of a cure opportunity violates due process and that cure deadlines that fall before voters can reasonably respond burden the right to vote. Several federal courts have agreed, ordering states to adopt or extend cure procedures. The legal trend is toward more cure, not less, which means the signature check is increasingly the beginning of a dialogue rather than a final judgment.

There is a security objection to the cure process, and it deserves a straight answer. The objection is that allowing voters to fix ballots after Election Day extends the window in which the system must remain trustworthy and creates an opportunity for bad actors to intervene. The answer administrators give is that the cure affidavit is itself a verified credential: it requires the voter’s signature under penalty of perjury, it is matched against the same registration record, and it is processed by the same bipartisan teams under the same observation rules as the original ballot. Nothing about the cure weakens the issuance log or the signature check. It adds a second, documented chance to satisfy them. The deeper objection is not really about security but about finality: elections are supposed to end, and every extension of the process extends the period of uncertainty. That is a genuine cost, and it is the reason cure deadlines are short and the affidavit requirements are strict. The system is buying accuracy with time, and like every trade in election administration, the price is visible.

Chain of Custody: Who Handles the Ballot

Chain of custody is the least visible part of mail voting and the part the skeptics scrutinize most closely, because it answers the question of who touches the ballot between the voter and the count. The documented chain begins when the voter relinquishes the ballot, to the postal service, to a drop box, or to an election office, and from that point every transfer is supposed to be logged. Postal carriers move the ballot in the ordinary mail stream, where it is subject to the same handling as any other mail, with the addition in many jurisdictions of the Intelligent Mail barcode that lets the election office see when the ballot entered the mail stream and when it arrived. Drop boxes are emptied on a schedule by election workers, and the collection is logged with the time, the box location, and the number of envelopes retrieved. Inside the election office, ballots move between intake, verification, opening, and tabulation stations under the observation of partisan observers, who are entitled in every state to watch the processing, and in many states the entire operation is recorded on video.

The standard the chain is held to is not perfection but documentation: every handoff should leave a record, so that a missing ballot can be traced to the point where the record stops. This is the same standard applied to evidence in a courtroom and to currency in a bank, and it is the reason administrators speak of the chain as a property of the records rather than of the physical ballot. A ballot that arrives without a matching issuance record is rejected. An envelope logged as received but never verified is flagged in the reconciliation. A batch of ballots counted without a corresponding batch of accepted envelopes is an anomaly that the canvass is designed to catch. The canvass, the official accounting that precedes certification, reconciles the number of voters credited with voting against the number of ballots counted, and a discrepancy must be explained before the election is certified. The chain of custody is therefore not a single safeguard but the connective tissue between all the others: it is what lets the issuance log, the signature check, and the tabulation corroborate one another.

The weak points in the chain are the ones the critics name, and they are real enough to describe honestly. The postal service loses mail, including ballots, at a low but nonzero rate, and a ballot lost in the mail is a vote that never arrives, with no remedy except the in-person fallback. Drop boxes, for all their surveillance, are unattended receptacles in public places, and isolated incidents of vandalism and arson have occurred. The collection rules that let third parties return ballots for others create a supervised-by-whom problem: in states that permit broad collection, a ballot can pass through hands the system never records. And the temporary workforce that processes millions of envelopes in a few weeks is trained quickly and supervised closely, but it is still a large number of people handling a large number of ballots under time pressure. None of these weak points is a secret. They are the reason the chain is documented the way it is, and the documented history of failures, which the evidence sections report in full, is the measure of whether the documentation has been enough.

How Mailed Ballots Are Counted

The counting of mailed ballots is the most ordinary part of the system, which is precisely why it is the least disputed. Once a ballot has been accepted and separated from its envelope, it is a paper ballot like any other, and it is counted on the same equipment, under the same observation rules, as ballots cast in person. Most jurisdictions use optical-scan tabulators: the voter marks an oval or completes an arrow, the scanner reads the marks, and the machine tallies the votes. Central-count jurisdictions feed accepted mail ballots through high-speed scanners at the election office. The tabulation equipment is tested before the election in public logic and accuracy tests, in which officials run a known set of test ballots through the machines and verify that the count matches, and the machines are sealed and secured between testing and use. Partisan observers are entitled to watch the counting, and in many jurisdictions the observation area is a permanent feature of the election office during the canvass period.

The timing of the count is where mail voting visibly differs from in-person voting, and the difference is a product of state law rather than technology. Before a mailed ballot can be counted, its envelope must be verified, and states differ on when that verification, called pre-processing or pre-canvassing, may begin. Some states allow offices to begin verifying and even scanning ballots weeks before Election Day, holding the results unreported until the polls close. Others forbid any processing until Election Day itself, which means millions of envelopes must be verified and opened while the in-person votes are also being counted. The states in the second group are the ones where mailed ballots are still being counted days after the election, and the delay is routinely misread as evidence of malfunction. It is not malfunction. It is the predictable result of a legal choice to compress all the processing into the shortest possible window. Whether that choice is wise is a legitimate subject of debate, and several states have moved their pre-processing dates earlier after experiencing the backlog, but the backlog itself is a scheduling artifact, not a sign of fraud.

The results from mailed ballots are reported alongside the in-person results, and the order in which they are reported has acquired a political significance it does not deserve. Because mailed ballots in many states are counted centrally and reported in large batches, and because different groups of voters use mail voting at different rates, the reported totals can shift as the mail batches are added to the in-person totals. This shift is an artifact of which ballots are counted first, not of anything happening to the ballots. Election administrators have a name for the misreading of this artifact, and they have spent years explaining that a lead which changes as more ballots are counted is the normal behavior of a count, not evidence of manipulation. The explanation has not always landed, which is itself a datum about the trust problem that the closing sections address. The machinery counts what it is given. The question of whether the public believes the count is a different question, and it is not answered by the tabulator.

Audits and the Paper Trail

What Does a Post-Election Audit Check in a Mail Ballot System?

A post-election audit re-examines a sample of the paper ballots, including mailed ones, to verify the tabulation. Auditors hand-count the sample and compare it to the machine count, expanding to more ballots on a mismatch. The audit checks the machines, not the voters: it confirms accepted ballots were counted as marked.

Mail voting has one structural advantage in the audit, which is that every vote exists on paper from the start. In-person voting on electronic machines without a paper record creates an audit problem: there is nothing to recount except the machine’s own memory. A mailed ballot is a physical artifact with the voter’s marks on it, which means the audit can always go back to the source. This is why election security specialists, even those critical of other aspects of mail voting, generally regard the paper trail as the system’s strongest feature. The paper does not depend on software behaving correctly. It can be examined by humans, stored under seal, and recounted years later if a court orders it. The retention rules require offices to keep ballots and envelopes for a fixed period, commonly 22 months for federal elections, which means the evidence of what the voters marked outlives the controversy about whether the machines read it correctly.

The most rigorous form of audit now in use is the risk-limiting audit, a statistical procedure that examines enough ballots to confirm the reported outcome with a stated level of confidence, and examines more ballots automatically when the margin is close. Colorado conducted the first statewide risk-limiting audits in 2017, and the method has since been adopted or piloted in a growing number of states. The risk-limiting audit is well suited to mail voting because it draws its sample from the paper ballots themselves, and mailed ballots, being centrally stored after processing, are straightforward to sample. A fixed-percentage audit, the older method still used in many states, hand-counts a set share of precincts or batches regardless of the margin, which is simpler to administer but less efficient. Both methods share the same purpose: to check the tabulation, not to re-verify the voters. Signature re-verification is a separate procedure, used in recounts and investigations, not in routine audits. The audit answers one question, did the machines count the accepted ballots correctly, and the mail voting pipeline, by producing a complete set of paper ballots, gives the audit the material it needs to answer it.

The Ballot’s Journey at a Glance

The six stages of the pipeline can be read as a single mechanism, and the table below compresses that mechanism into the form a reader can check any claim against. Each row is one stage of the ballot’s journey. The columns show what happens at that stage, which verification is applied, and the failure mode the verification is designed to catch. The table settles a specific question: where, exactly, could something go wrong with a mailed ballot, and what in the system is positioned to stop it. A claim about mail voting fraud that cannot be located in one of these rows is a claim about a part of the process that does not exist.

Stage	What Happens	Verification Applied	Documented Failure Mode
Issuance	The office mails or hands the ballot packet to the voter	Request checked against the registration record; every ballot logged	Ballot sent to a stale address; wrong ballot style issued
Return	The voter marks, seals, signs, and returns the ballot	None at this stage; the envelope carries the credential	Ballot lost in transit; ballot collected by an unauthorized third party
Verification	Workers check the envelope before opening it	Signature compared to registration record; issuance log checked	Valid signature wrongly rejected; invalid signature wrongly accepted
Custody	Accepted ballots are stored and moved to tabulation	Handoff logs; bipartisan observation; sealed storage	Batch misplaced; seal broken without documentation
Tabulation	Scanners read the marked ballots and tally votes	Logic and accuracy testing; observer access; canvass reconciliation	Machine misread; batch counted twice
Audit	A sample of paper ballots is hand-counted and compared	Statistical sampling; expanded examination on mismatch	Discrepancy found too late to affect certification

The table is worth returning to when the evidence sections discuss fraud, because every documented case of mail ballot fraud maps onto one of these rows, and the pattern of which rows produce cases and which do not is itself informative. The rows that generate the most documented cases are return and verification: ballots intercepted or collected by third parties, and signatures that should have been caught. The rows that generate the fewest are tabulation and audit: the counting machinery, once the ballot is accepted, has produced vanishingly few documented failures. That distribution is the empirical shape of the system’s risk, and it is why administrators concentrate their reforms on the front of the pipeline, on issuance accuracy and verification quality, rather than on the counting equipment at the end.

The Federal Frame I: Overseas and Military Voters

The oldest continuous mail voting system in the United States is the one Congress built for voters who cannot reach a polling place because they serve abroad or live outside the country. The Uniformed and Overseas Citizens Absentee Voting Act of 1986 requires every state to allow members of the uniformed services, their eligible family members, and overseas civilians to register and vote absentee in federal elections. The Military and Overseas Voter Empowerment Act of 2009 strengthened those protections by requiring states to transmit requested ballots at least 45 days before a federal election and to offer electronic transmission of blank ballots, though the voted ballot itself must still come back by mail in most states. The federal write-in absentee ballot serves as a backup: a voter who requested a ballot but never received one in time may cast this standardized federal ballot, and it is counted for federal offices if the state ballot never arrives.

The overseas system is instructive because it operates under the hardest conditions mail voting faces: the longest transit times, the least reliable postal networks, and voters scattered across time zones. Its failure modes are well documented. Ballots mailed too late to reach far-flung voters, ballots returned too late to meet state deadlines, and ballots rejected over signature or witness technicalities have disenfranchised service members in measurable numbers across many elections, which is why Congress kept tightening the transmission deadlines. The system’s defenders point out that it enfranchises millions of voters who would otherwise have no practical way to vote, and that its problems are problems of logistics rather than of integrity. No serious account of mail voting fraud centers on the overseas pipeline, because the population is small, the ballots are closely tracked, and the voters are identifiable. The overseas system demonstrates the basic bargain of mail voting in its purest form: the ballot travels far, the safeguards travel with it as procedure, and the cost of the distance is paid in rejected and late ballots rather than in compromised ones.

The overseas framework also illustrates the limits of federal power over election administration. Congress can set the transmission deadline and require the states to offer the federal write-in ballot, but it cannot dictate how states verify the returned ballots or when they count them. Those decisions remain with the states, which is why an overseas ballot mailed from the same country can face different verification rules depending on the voter’s state of residence. This federal-state division runs through all of election law, and it is the reason mail voting looks like fifty different systems rather than one. The federal government sets floors, through the statutes described in the next section, and the states build their own structures on top of them. Any account of how vote-by-mail works that describes only the federal rules, or only one state’s rules, is describing a fraction of the system.

The Federal Frame II: HAVA and the Federal Floors

Two federal statutes set the floors under every state mail voting system, and both were written in response to failures that the country decided not to repeat. The Help America Vote Act of 2002, passed after the disputed presidential election of 2000, required every state to maintain a single statewide computerized voter registration list, set minimum standards for voting equipment, required provisional ballots for voters whose eligibility is uncertain, and imposed identification requirements on first-time voters who registered by mail. The identification rule is the provision that touches mail voting most directly: a first-time voter who registered by mail must provide identifying information, either when voting in person or with the mailed ballot, before the ballot can be counted. The statute’s full requirements, including its equipment standards and its funding of state election administration, are covered in the companion guide to the Help America Vote Act. What matters here is the principle: Congress used its authority over federal elections to require that the registration list behind the mailed ballots be statewide, computerized, and maintained to a standard, which is the precondition for the issuance log to work at all.

The National Voter Registration Act of 1993 set the other floor, governing how voters get onto the rolls in the first place. By requiring motor vehicle offices, mail registration, and public assistance agencies to offer registration, and by restricting purges, the statute determines the size and composition of the list from which mail ballots are issued. States that mail ballots to every registered voter are, in effect, mailing ballots to the list the registration act built. The statute’s mechanics, including its limits on list maintenance and the litigation those limits produced, are covered in the companion guide to the registration law. The interaction between the two statutes is what makes the federal frame coherent: one governs entry onto the rolls, the other governs the integrity of the rolls themselves, and the mail voting pipeline assumes both are functioning. When either fails, when registrations are missing or the list is stale, the mail system inherits the failure, and no amount of envelope procedure can fix a ballot mailed to the wrong address because the registration behind it was wrong.

What the federal government has not done is standardize mail voting itself. There is no federal statute prescribing how states verify signatures, when they may begin processing ballots, or whether they must offer drop boxes or cure procedures. Bills to set national standards have been introduced in many Congresses and have not passed, which means the fifty-state variation is not a temporary condition awaiting federal harmonization but the durable shape of American election administration. Courts have filled some of the gaps, ordering cure procedures or striking down specific restrictions, but court orders are piecemeal by nature. The result is a system in which the federal floors guarantee certain minimums, statewide registration lists, identification rules for first-time mail registrants, ballot access for overseas voters, and the states decide everything else. A reader who understands this division understands why mail voting can be simultaneously well-regulated and wildly variable: the regulation is real, and so is the variation.

The First Absentee Voters: The Civil War

Mail voting in the United States began as a wartime necessity, and its origin explains why the practice has always been tied to the question of who deserves access to the ballot. During the Civil War, hundreds of thousands of Union soldiers served far from home, and their absence threatened to disenfranchise the very men the nation was asking to fight. Beginning in 1862, Northern states passed soldier-voting laws allowing troops to cast ballots from the field, either by mail or through in-person voting at military camps supervised by state commissioners. The details varied by state. Some allowed soldiers to mail ballots home. Others sent commissioners to collect votes at the front. Pennsylvania’s 1864 soldier-voting law survived a challenge in the state supreme court, which upheld the legislature’s power to provide for voting outside the voter’s district. The election of 1864, in which the soldier vote ran heavily for Abraham Lincoln, demonstrated that absentee voting could operate at national scale without collapsing, and the practice never fully disappeared after the war ended.

The Civil War precedent established two patterns that have persisted. The first is that expansions of absentee voting follow the enfranchisement of groups whose circumstances keep them from the polls: soldiers then, and later the elderly, the disabled, travelers, and eventually any voter who prefers the convenience. The second is that every expansion is contested on the same grounds, that voting outside the supervised polling place invites fraud and coercion. The contest over soldier voting in the 1860s featured accusations that officers pressured enlisted men and that ballots were manipulated in transit, accusations that echo in the modern debate over ballot collection. The historical record does not resolve whether those accusations were well founded, but it establishes that the argument about mail voting is older than the telephone. Every generation rediscovers it, because the underlying tension, between making voting accessible and keeping it supervised, is structural rather than partisan. The technology changes. The tension does not.

After the war, absentee voting contracted and then grew again in waves. States extended absentee provisions to civilians in the late nineteenth and early twentieth centuries, usually with excuse requirements that kept the practice narrow. The twentieth century’s great expansion came in stages: the extension to overseas civilians, the gradual adoption of no-excuse absentee voting beginning in the 1970s and 1980s, and then the universal systems of the late 1990s onward. Each wave followed the same political logic. The proponents argued that the existing rules excluded legitimate voters for no good reason. The opponents argued that each relaxation removed a safeguard. The proponents generally won, slowly, state by state, which is why the map of mail voting in the United States looks like an accretion of local decisions rather than a designed system. It is an accretion of local decisions, and understanding any particular state’s rules requires knowing which wave put them there.

The Modern Invention: Oregon and the All-Mail States

The universal vote-by-mail system, in which every registered voter automatically receives a ballot, was invented in Oregon in the 1990s, and its history is the clearest natural experiment mail voting has produced. Oregon began experimenting with all-mail elections for special elections in the late 1980s, and in 1998 the voters approved Ballot Measure 60, making Oregon the first state to conduct general elections entirely by mail. The measure passed with a large majority, and the state has run every election by mail since. Washington followed, with counties adopting all-mail voting through the 2000s and the state completing the transition in 2011. Colorado’s 2013 Voter Access and Modernized Elections Act moved the state to all-mail elections while adding same-day registration and in-person voting centers. Utah’s counties adopted all-mail voting through the 2010s, and the state formalized universal mail elections for 2020. Hawaii moved to all-mail voting beginning with the 2020 elections. California, Vermont, and Nevada made universal mail voting permanent in 2021. Eight states, built over two decades, one decision at a time.

Oregon’s long record is the evidence base that researchers return to whenever they need a jurisdiction with decades of all-mail experience. Turnout in Oregon has consistently run above the national average, though researchers debate how much of the difference the mail system explains and how much reflects the state’s political culture and demographics. Documented fraud in Oregon’s all-mail elections has been minimal across more than two decades and tens of millions of ballots cast, a record that defenders of mail voting cite as the strongest available evidence that the system can operate securely at scale. The state’s administration has also produced the most mature version of the procedures described above: centralized ballot tracking, signature verification with cure, drop boxes statewide, and public reporting of ballot status. Oregon did not merely adopt mail voting. It debugged it, in public, over twenty years, and the procedures other states now copy are in large part the procedures Oregon developed through trial and error.

The spread of the universal model also reveals the politics of mail voting adoption, which has not followed a simple partisan pattern. Oregon’s measure passed with bipartisan support in a state that was then politically competitive. Utah, one of the most Republican states in the country, adopted all-mail voting through county-level decisions driven by administrators who found it cheaper and more reliable than running polling places across a rural state. Colorado’s reform was enacted by a Democratic legislature over Republican objections, and the state’s subsequent elections have been administered without the fraud its opponents predicted. California’s permanent adoption came from a Democratic legislature during the pandemic. The pattern is that administrators who run all-mail elections tend to defend them regardless of party, while the fiercest opposition comes from politicians who have never administered one. That pattern does not prove the system is flawless. It suggests that familiarity with the machinery changes how people assess it, which is consistent with the trust-gap argument developed in the closing sections.

The 2020 Expansion and What It Endured

The pandemic election of 2020 produced the largest natural experiment in mail voting history, and its results are the empirical foundation of the modern debate. Facing a virus that made crowded polling places dangerous, states expanded mail voting farther and faster than ever before: states that had never offered no-excuse absentee voting adopted it, some temporarily and some permanently; offices mailed ballot applications, and in a few states ballots themselves, to every registered voter; and the share of Americans voting by mail roughly doubled compared with 2016, reaching about four in ten voters. The expansion was improvised under emergency conditions, administered by offices that had months rather than years to scale up, and conducted through a postal service that was itself under strain. If mail voting had a fundamental, structural vulnerability to fraud at scale, the conditions of 2020 were the ones most likely to expose it.

What the record shows is that the system held. The election produced the highest turnout in more than a century, the count survived more than sixty lawsuits challenging the results, and the audits and recounts conducted afterward, including hand recounts in several closely contested states, confirmed the reported outcomes. The documented fraud was real but small: scattered cases of individuals voting twice, voting on behalf of deceased relatives, or attempting to cast ballots they were not entitled to, numbering in the hundreds across more than 150 million votes cast. Researchers who studied the expansion found that it raised turnout, particularly among groups that vote less often, without systematically benefiting either party. The most careful studies compared counties and states that expanded mail voting against similar ones that did not, and the consistent finding was a modest turnout increase and no partisan effect. The expansion did not prove that mail voting is fraud-proof. No election administrator claims that. It demonstrated that a rapid, large-scale expansion of mail voting, conducted under adverse conditions, produced an election whose results withstood every form of scrutiny the system provides.

The aftermath of 2020 reshaped the map in both directions, and the reshaping is now part of the system’s permanent history. Several states made their pandemic-era expansions permanent, joining or approaching the universal model. Others restricted mail voting, tightening identification requirements, limiting drop boxes, shortening request windows, or banning third-party ballot collection. The restrictions were generally defended as security measures and attacked as voter suppression, and the courts have upheld some and struck down others, in fights that play out against the federal voting-rights framework traced in the history of Voting Rights Act reauthorizations. What did not happen, in either direction, was a collapse of the underlying machinery. States that expanded mail voting continued to verify signatures, reconcile issuance logs, and audit counts. States that restricted it continued to process the mail ballots they still allowed. The system proved adaptable in both directions, which is evidence about the machinery’s robustness rather than about the wisdom of any particular rule change. The rules will keep changing. The pipeline they govern has now been tested at a scale no one had attempted before.

How Other Democracies Handle Postal Voting

The United States is not the only democracy that lets voters cast ballots by mail, and the comparison clarifies which features of the American system are choices and which are necessities. Germany has allowed postal voting, called Briefwahl, since 1957, originally for voters unable to reach the polls and since 2008 without any excuse required. In the 2021 federal election, nearly half of German voters cast their ballots by mail, a share comparable to the American peak. The German system mails ballot materials automatically to voters who request them, requires the voter to sign a declaration on the return envelope affirming that the ballot was marked personally or with the assistance of a named helper, and counts postal ballots alongside in-person ones at the polling-station level, with the envelopes opened and the ballots mixed before counting to preserve secrecy. Documented fraud has been rare, and the system’s legitimacy is broadly accepted across the German political spectrum.

Britain took a different path to a similar destination. The Representation of the People Act 2000 introduced postal voting on demand in England, Scotland, and Wales, removing the excuse requirement that had previously limited the practice. Any registered voter may now request a postal ballot, and the completed ballot must be returned by the close of polls, with the voter’s date of birth and signature on the return statement checked against the application record. Britain’s experience includes the cautionary episode that American skeptics cite most often from abroad: the 2004 Birmingham local elections, in which a judge found that postal ballot fraud had been widespread enough to void results in two wards, describing the system as one that would disgrace a banana republic. The episode led to reforms, including the requirement that postal voters provide personal identifiers, and subsequent elections have not repeated the failure. The British record thus contains both the strongest foreign evidence that postal voting can fail and the evidence that the failure was diagnosable and fixable.

Switzerland offers the third instructive case, a country where postal voting is not an accommodation but the norm. Federal law has permitted postal voting in national referendums and elections since 1994, following earlier cantonal experiments, and the overwhelming majority of Swiss votes are now cast by mail. The Swiss system works because the country’s frequent referendums made in-person voting for every question impractical, and the population adapted to a rhythm of receiving ballot materials, marking them at home, and returning them by mail or drop-off. Fraud is rare and the system’s legitimacy is unquestioned, which demonstrates that a polity can conduct essentially all of its voting by mail without the practice becoming controversial. The Swiss case also shows what the American debate sometimes misses: where postal voting is universal and long-standing, it ceases to be a partisan question at all, because every party’s voters use it and every party’s administrators run it. Controversy attaches to the change, not to the method.

What Those Systems Chose Differently

The three foreign systems differ from the American one in design choices that illuminate the American tradeoffs. Germany counts postal ballots at the polling-station level rather than centrally, which means the ballots are integrated into the local count on election night and the results carry no visible mail-versus-in-person split. The American practice of central counting, by contrast, produces the batch-reporting artifact in which mail totals arrive separately and shift the reported lead, feeding the misreading described earlier. Germany also requires the signed declaration of personal marking, a legal attestation that substitutes for the American signature-verification infrastructure. The German choice trusts the voter’s sworn statement more and the administrator’s handwriting comparison less. Which choice is better depends on what failure one fears more: impersonation, which the American signature check targets, or coercion, which neither system can fully prevent and which the German declaration at least forces the voter to attest against.

Britain’s identifier requirement, the date of birth and signature on the postal voting statement, represents a middle position between the American signature check and the German declaration. The British system checks two identifiers rather than one, which reduces the reliance on handwriting comparison, but it also rejects ballots where the voter misremembers or miswrites the date of birth, a failure mode the Americans avoid. The British reforms after the Birmingham episode also illustrate a principle the American system follows unevenly: when a failure is documented, the response is to add a specific check aimed at the documented failure rather than to restrict the method as a whole. Britain did not abolish postal voting after 2004. It added identifiers, tightened the application process, and gave returning officers clearer powers. The American states that restricted mail voting after 2020 sometimes followed the same logic and sometimes restricted the method itself, and the difference between those two responses is the difference between engineering and politics.

Switzerland’s lesson is the most basic and the hardest to transfer. The Swiss system works smoothly in part because Swiss elections are administratively simple: frequent referendums with short ballots, a small population, and a postal service that delivers reliably. American ballots are long and complex, American elections are administered by thousands of county offices with different equipment and procedures, and the American postal network, while extensive, is not designed around election deadlines. None of this means the Swiss experience is irrelevant. It means the American system faces harder engineering problems than the Swiss one, and its procedures, signature verification, issuance logs, pre-processing rules, are the engineering answers to those harder problems. A system that looks overcomplicated is often a system solving a complicated problem. The comparison does not flatter or condemn the American design. It explains why the design looks the way it does.

The Documented Failure: North Carolina’s Ninth District

Has Mail Ballot Fraud Ever Changed an Election Outcome?

Yes, once in recent American history. In the 2018 race for North Carolina’s Ninth Congressional District, operatives collected absentee ballots and investigators found evidence some were altered after collection. The State Board of Elections voted unanimously in February 2019 to invalidate the election and order a new one, which the system’s own records had made possible.

The North Carolina case deserves careful study because it shows both the vulnerability the skeptics warn about and the detection the defenders rely on. The vulnerability was in the return stage of the pipeline, the least supervised link: Dowless’s workers collected ballots directly from voters, which was legal under North Carolina law at the time for the voters’ designated agents, and the operation exploited the gap between what the law permitted and what it could supervise. Red flags accumulated in the data before anyone proved fraud: an unusually high rate of absentee ballot requests in Bladen County, a pattern of unreturned ballots concentrated in certain precincts, and witness testimony about workers visiting voters’ homes. The detection came through the system’s own records: the issuance and return logs showed the anomalous patterns, voters testified about what the collectors had done, and the board’s investigators followed the paper trail. The election was not stolen and certified. It was investigated and voided, and the remedy, a new election, was the system working as designed, at the cost of months of uncertainty and a second campaign.

The aftermath changed the law in ways that illustrate how the system learns from failure. North Carolina tightened its absentee ballot procedures, requiring additional information on request forms and strengthening the rules on who may handle ballots. Other states cited the case when writing or tightening their own ballot collection laws. The case also clarified the evidentiary standard the skeptics must meet: it is not enough to show that fraud is possible, or that the chain of custody has gaps. The North Carolina operation was possible, it exploited real gaps, and it was caught, investigated in public, and reversed. A critic who argues that similar operations are occurring undetected elsewhere is making a claim the North Carolina record cuts both ways on: the same machinery that caught Dowless exists in every state, in the form of issuance logs, return patterns, and canvass reconciliation, and the absence of comparable findings elsewhere is evidence, though not proof, that comparable operations are not widespread. The honest reading of the case is that mail voting’s weakest link is real, that the system’s immune response is also real, and that both facts have to be held at once.

The case also carries a warning about scale that both sides sometimes miss. The Dowless operation targeted a single congressional district and involved hundreds of ballots, and it required a paid network of workers visiting voters’ homes over weeks. Scaling such an operation to affect a statewide or national election would require replicating it across many counties, each with its own election office keeping its own logs, which multiplies both the cost and the probability of detection. This is not an argument that large-scale fraud is impossible. It is an argument about what large-scale fraud would have to look like: not a single clever hack but a vast, visible conspiracy of thousands of participants, every one of them a potential witness, operating against a system whose records are designed to reveal exactly the anomalies such an operation produces. The North Carolina conspirators were caught by a handful of statistical anomalies in one county’s returns. A statewide operation would produce anomalies in every county, and the investigators would not need to look hard to find them.

The Chain-of-Custody Argument and Its Evidence

The strongest intellectual case against mail voting does not rest on any single fraud, but on a structural claim: that a ballot marked outside a supervised polling place passes through hands the system cannot fully account for, and that this gap makes fraud easier to attempt and harder to detect than in-person impersonation. The claim’s leading advocates include the election-integrity researchers associated with the Heritage Foundation, whose election fraud database compiles proven cases across all voting methods, and commentators such as Hans von Spakovsky, who argue that the chain-of-custody gap is the central vulnerability of mail voting. Their argument has a clear logic. In-person voting authenticates the voter at the moment of voting, under observation. Mail voting authenticates the voter after the fact, by comparing a signature on an envelope that has traveled through the mail. Between the voter’s hand and the verifier’s desk lies a gap, and the argument is that a gap is a gap: whatever the statistics say about detected fraud, undetected fraud in the gap cannot be measured, so the statistics understate the risk.

The argument deserves its best form before it is answered, because it identifies a real asymmetry. In-person voter impersonation is hard to commit at scale: it requires a fraudster to appear at a polling place, claim to be someone else, and do so repeatedly without being recognized, for a payoff of one vote per appearance. Mail ballot fraud, by contrast, can be committed from a desk: a fraudster who obtains blank or completed ballots can mark and return many of them, which is why the documented cases of mail fraud, though few, tend to involve larger numbers of ballots per incident than the documented cases of in-person fraud. The Heritage database, for all its limitations as a comprehensive measure, consistently shows mail-related cases forming a disproportionate share of proven election fraud relative to mail’s share of votes cast. The critics’ inference is that the system’s detection machinery is weaker than its defenders claim, and that the absence of evidence is not evidence of absence.

The defenders’ answer operates on three levels, and each level concedes something. First, they concede the asymmetry but dispute its significance, pointing out that the fraud that is easiest to commit at scale is also the easiest to detect at scale, because the issuance log, the return patterns, and the signature check are designed to catch exactly the anomalies that batch fraud produces. The North Carolina case is their exhibit: the fraud was attempted at the return stage, and the system’s records caught it. Second, they point to the base rate: even if undetected fraud exists, the detected rate is so low, measured in hundredths of a percent of ballots cast, that undetected fraud would have to be orders of magnitude larger than detected fraud to change outcomes, and there is no mechanism proposed by which such a large hidden enterprise could operate without producing the anomalies the system is built to notice. Third, they note that the critics’ own evidence base, the Heritage database, documents hundreds of cases across billions of ballots cast over decades, which is consistent with a rare-crime account rather than a systemic-vulnerability account. A database assembled by the critics, showing small numbers, is stronger evidence for the defenders than any study the defenders could commission themselves.

What remains after the exchange is a narrower disagreement than the public debate suggests. Both sides agree that the return stage is the weakest link. Both sides agree that the documented fraud rate is low. The critics believe the low rate reflects weak detection; the defenders believe it reflects rare commission. That belief cannot be settled by more statistics, because it is a belief about the unobserved, and beliefs about the unobserved are settled, if they are settled at all, by institutional design rather than by data. The defenders’ final move is therefore institutional: make the detection machinery stronger, publish the rejection and cure statistics, open the processing to observation, and let the system’s transparency carry the burden the statistics cannot. Whether that burden can be carried is the trust question, and it is taken up in the closing sections.

Rejection Rates and Who Bears Them

If fraud is the risk the skeptics emphasize, rejection is the risk the system’s own data emphasizes, because rejected ballots are counted, published, and attributable. The national rejection rate for mail ballots in federal elections runs around one percent, which translates, at the scale of tens of millions of mailed ballots, into hundreds of thousands of voters whose ballots are set aside. The leading causes are consistent across states and elections: missing signatures, signatures that do not match the registration record, and ballots that arrive after the deadline. Late arrival is the largest single cause in most jurisdictions, which is why administrators plead with voters to return ballots early. Signature problems are the second largest, and they are the ones that raise the fairness question, because they do not fall evenly.

Research on mail ballot rejection has found a persistent pattern: rejection rates are higher for young voters, for first-time mail voters, and for Black and Hispanic voters than for older white voters. Studies of Florida’s elections by Enrijeta Shino, Mara Suttmann-Lea, and Daniel Smith documented rejection rates near one and a half percent overall, with the burden falling disproportionately on voters whose signatures were least stable and whose familiarity with the cure process was thinnest. The mechanisms are not mysterious. Young voters’ signatures change more as they age into adulthood. Voters who registered long ago have reference signatures that no longer resemble their current hand. Voters with less experience of the mail system are less likely to know that a cure affidavit exists or that the deadline for returning it is short. None of this requires discriminatory intent by election workers. It requires only a verification technology, human signature comparison, whose error rate correlates with the voter’s age and experience, applied uniformly across a diverse electorate.

The policy response has concentrated on the cure process and on voter education, and the evidence suggests both help. States that notify voters quickly and through multiple channels, and that give voters several days after Election Day to cure, recover a substantial share of initially rejected ballots. Voter education campaigns that show voters where to sign, warn them about deadlines, and explain the tracking systems reduce the late-arrival and missing-signature rejections that need no judgment to prevent. Some administrators have gone further, proposing to supplement or replace signature verification with other credentials, though every proposed alternative trades one set of problems for another, as the earlier section on verification discussed. The deeper point is that rejection is the visible cost of the system’s security architecture, and it is a cost the system can measure and reduce. A voting method whose failures are countable is a voting method that can be held accountable, and the rejection statistics, uncomfortable as their distribution is, are the accountability mechanism working.

There is a final dimension to the rejection question that the fraud debate tends to obscure. A rejected mail ballot is not the same as a suppressed vote, because the voter who learns of the rejection in time can cure it, and the voter who never mailed the ballot can still vote in person. The system’s fallbacks mean that rejection is usually remediable, which is why administrators distinguish between the initial rejection rate and the final count of ballots that were never counted. The distinction matters for how the evidence is read. A one percent initial rejection rate with a robust cure process produces a far smaller final disenfranchisement rate than the same initial rate with no cure at all. This is also why the cure litigation has been so consequential: every court order extending cure rights or deadlines directly reduces the number of voters who lose their vote to a signature mismatch. The system’s harshest critics and its most aggressive reformers agree on one thing, though they would phrase it differently: the signature check is where the system’s promise and its practice diverge most visibly, and it is where the next decade of reform will concentrate.

The Turnout Evidence

Does Voting by Mail Increase Turnout?

Yes, modestly. The most careful studies find that universal vote-by-mail raises turnout by roughly two to four percentage points, concentrated among infrequent voters. Request-based systems show smaller, less consistent effects. The broader literature centers near two points in presidential years, and no credible study finds that mail voting reduces turnout.

The turnout literature is unusually well developed for an election administration question, because the staggered adoption of mail voting across states created the natural experiments researchers need. The foundational study, by Alan Gerber, Gregory Huber, and Seth Hill, examined Washington State’s transition to all-mail elections and found turnout increases of two to four percentage points, with the largest gains among occasional voters. A 2020 study by Daniel Thompson, Jennifer Wu, Jesse Yoder, and Andrew Hall, published in Science Advances, used county-level comparisons across states that adopted universal vote-by-mail and found turnout gains of roughly two to three percentage points, a result that held across the political spectrum of the adopting states. A 2009 study prepared for the Pew Charitable Trusts found that no-excuse absentee voting raised participation by about three percentage points relative to excuse-required systems. The consistency of the magnitude, two to four points across different states, different decades, and different research designs, is what gives the finding its credibility. It is not one study’s result. It is the center of a distribution of results.

The mechanism behind the turnout effect is convenience, but convenience understood precisely. Mail voting does not make nonvoters into voters by inspiration. It lowers the cost of voting for people who are already inclined to participate but face logistical obstacles: the shift worker who cannot reach the polls during voting hours, the parent without childcare, the rural voter miles from the nearest polling place, the voter with a disability for whom travel is difficult. The ballot arrives without a trip, it can be marked over days rather than minutes, and it can be returned without waiting in line. Researchers describe this as reducing the transaction cost of voting, and the voters whose behavior changes most are those for whom the transaction cost was the binding constraint. This is also why the effect is larger in low-salience elections, where the motivation to overcome obstacles is weaker and convenience matters more, and smaller in presidential elections, where motivation is already high. The 2020 expansion confirmed the pattern: turnout rose across the board, but the mail voting component of the increase was concentrated where the literature predicted.

Two caveats keep the turnout finding honest. The first is selection: the states that adopted universal mail voting may differ from other states in ways that also affect turnout, such as political culture or civic infrastructure, and researchers address this with the county-level and within-state comparisons described above, but no observational design eliminates the concern entirely. The second is the distinction between the universal and request-based systems: most of the strong turnout evidence comes from universal systems, and the evidence for request-based no-excuse absentee voting is weaker and more contested, with some studies finding little or no effect once the composition of who requests ballots is accounted for. The responsible summary is therefore conditional: universal mail voting reliably produces a modest turnout increase; no-excuse absentee voting probably produces a smaller one; and the effect in both cases runs through convenience rather than persuasion. A reader who encounters a claim that mail voting transforms participation, in either direction, is encountering a claim the evidence does not support.

The Partisan Question

The most politically charged question about mail voting is whether it helps one party, and the evidence gives a clear answer that neither party’s strategists fully trust. The consistent finding across the major studies is that mail voting does not systematically advantage Democrats or Republicans. The Thompson study in Science Advances tested the partisan effect directly and found no measurable impact on either party’s vote share. Earlier work on Washington, Oregon, and Utah, states spanning the political spectrum, found turnout increases distributed across party registration roughly in proportion to each party’s share of the electorate. The MIT Election Lab’s synthesis reports the same null result. The mechanism is intuitive once stated: convenience helps whichever voters face the highest transaction costs, and those voters are not concentrated in one party. Rural Republican voters far from polling places benefit alongside urban Democratic voters with inflexible work schedules, and the net effect washes out.

The history of party positions on mail voting is itself evidence for the null finding, because the parties have switched sides as their coalitions changed. Through the late twentieth century, Republicans were often the stronger supporters of absentee voting expansion: their voters skewed older, and older voters used absentee ballots at high rates, while military voters, another Republican-leaning group, voted absentee by necessity. Democrats, whose coalition included more young and low-income voters, were relatively less invested in the practice. The alignment began shifting in the 2000s and 2010s as Democrats embraced mail voting as an access measure and some Republicans came to see it as a Democratic advantage, a perception the research does not support but that shaped party strategy anyway. By the 2020s, the parties’ public positions had substantially reversed from where they stood in the 1980s, while the underlying evidence on partisan effects had barely moved. Parties change their minds about election rules when they change their minds about who benefits, and the history suggests those judgments track perception more closely than measurement.

The null partisan finding carries an important implication for how the rest of the evidence is read. If mail voting systematically helped one party, then disputes about its rules would be straightforward conflicts of interest, and the security debate would be downstream of the power struggle. The evidence that it helps neither party does not make the disputes disinterested, because politicians act on beliefs about advantage whether or not the beliefs are correct, but it changes the character of the disagreement. A fight over the rules of a neutral procedure is a fight about values, about how much convenience a voting system should offer, how much risk it should tolerate, and how much trust it should demand, rather than a fight about who wins. That is a more honest frame for the controversy, and it is the frame the closing sections adopt. The procedure is neutral in its effects. The argument about the procedure is not neutral, because it is an argument about what voting should require of citizens, and that question has no technical answer.

The Strongest Case Against Mail Voting

The case against mail voting, stated at its strongest, is not that fraud is rampant but that the system asks citizens to trust what they cannot see. Its advocates begin with the polling place as the baseline: a supervised environment where the voter’s identity is checked in person, the ballot is marked in private but under observation, and the ballot goes directly from the voter’s hand into the tabulator or the ballot box. Every step is visible to the voter and to the observers the parties station at the polls. Mail voting replaces that visibility with procedure. The voter cannot watch the signature comparison. The voter cannot watch the envelope opened. The voter cannot watch the ballot scanned. The system offers transparency in the form of published procedures, observer access, and audit reports, but these are transparency for specialists, not for citizens. The ordinary voter is asked to take the system’s word for it, and the critics argue that a democratic procedure should not require that kind of faith.

The critics’ second argument concerns the home as a voting environment. The secret ballot was invented to protect voters from coercion, and the polling place enforces it architecturally: no one may accompany the voter into the booth, no one may watch the voter mark the ballot, and electioneering is banned within a set distance. The home enforces nothing. A domineering family member, an employer, a union steward, or a ballot collector can watch, pressure, or direct the voter’s choices, and the system has no way to know. Documented cases of such coercion are rare, but the critics argue, with some force, that coercion in the home is the hardest form of election misconduct to detect, because the victim has every reason not to report it and the system has no record that it occurred. The secret ballot at the polling place solves a problem that mail voting reintroduces, and no signature check or chain-of-custody log addresses it, because the violation happens before the envelope is ever sealed.

The third argument is the chain-of-custody gap developed at length above, and the critics’ version of it ends where the defenders’ version begins. The defenders say the gap is monitored by logs, checks, and audits. The critics reply that monitoring is not prevention, and that a system which detects fraud after the fact is a system that has already failed the voters whose ballots were compromised. They point to the rejection statistics as evidence that the verification machinery is imperfect, to the North Carolina case as evidence that the gap can be exploited, and to the impossibility of measuring undetected fraud as evidence that the published fraud rates are a floor rather than a ceiling. The cumulative claim is modest in its logic and sweeping in its implication: mail voting may be well administered, but it is administered on trust, and trust is not a safeguard. A voting system should be designed so that it does not need to be trusted, and mail voting, by moving the act of voting out of public view, fails that test in principle even when it passes it in practice.

The Strongest Case For Mail Voting

The case for mail voting, stated at its strongest, begins with the voter the polling place leaves out. Every requirement attached to in-person voting, travel to a specific location, availability during specific hours, physical ability to stand in line, freedom from work or caregiving obligations during those hours, filters the electorate. The filters do not fall randomly. They fall hardest on the hourly worker who cannot take time off, the single parent without childcare, the rural voter an hour from the polling place, the voter with a disability, the student away at school. Mail voting removes those filters at a stroke. The ballot comes to the voter, the voter marks it on their own schedule, and the return options, mail, drop box, office delivery, accommodate every circumstance. The defenders argue that a voting system should be judged by how many eligible citizens it enables to vote, and that on this measure mail voting outperforms the polling place by the margin the turnout studies document. Access is not a side benefit of the system. It is the system’s reason for existing.

The defenders’ second argument is that the polling place’s supposed superiority is partly nostalgic. In-person voting has its own documented failure modes: long lines that deter voters in under-resourced precincts, equipment failures, poll worker errors, voter intimidation at the polls, and the same human fallibility in identity checking that the critics attribute to signature verification. The polling place is supervised, but supervision is not infallibility, and the history of American elections contains ample evidence of misconduct inside polling places, from ballot box stuffing in the nineteenth century to the voter suppression campaigns of the twentieth. The critics compare mail voting to an idealized polling place that has never existed. The defenders compare it to the actual alternative, which is a system with different vulnerabilities rather than no vulnerabilities. On that comparison, mail voting’s documented fraud rate, measured across billions of ballots, stands up well, and its paper trail gives auditors material that electronic in-person voting sometimes lacks.

The third argument is institutional and forward-looking. The defenders point to the jurisdictions with the longest mail voting experience, Oregon’s quarter century, Washington’s, Colorado’s, Utah’s, and observe that sustained administration produces sustained legitimacy. Election officials in those states, of both parties, defend their systems. Voters use them at high rates and report high satisfaction. Documented fraud remains minimal. The pattern suggests that the trust problem the critics identify is sharpest where mail voting is newest and most contested, and mildest where it is oldest and most routine. Familiarity, in other words, is the solvent of suspicion, which implies that the controversy is transitional rather than structural. The defenders’ strongest claim is therefore a prediction, though they would frame it as an observation: give the system time, administer it transparently, publish the data, and the argument will subside, not because the skeptics are refuted but because the voters who use the system will have learned, from experience, that it works.

The Complication: The Trust Gap

The honest complication for the thesis is that the thesis can be entirely correct about the machinery and still fail to persuade, because the dispute about mail voting is not, at bottom, a dispute about the machinery. Consider what the evidence establishes: the pipeline is documented, the fraud rate is low, the turnout effect is modest and nonpartisan, the failure modes are known, and the system has survived its largest stress test. A reader who accepts all of this has every reason to regard mail voting as a sound administrative practice. Yet the controversy persists, and it persists among people who have access to the same evidence. The thesis survives this observation only if it can explain it, and the explanation is that election legitimacy is not a property of the counting machinery. It is a property of the relationship between the machinery and the public, and that relationship can fail while the machinery works perfectly.

The trust gap has a structure worth describing precisely. Trust in an election requires the voter to believe three things: that eligible voters could vote, that ineligible votes were not counted, and that the count reflects the votes cast. Mail voting is strong on the first, measurably adequate on the second, and procedurally sound on the third, but it is weak on the voter’s ability to verify any of the three personally. The polling place lets the voter see the process. Mail voting asks the voter to believe the process. That difference is not a flaw in the voter’s reasoning. It is a real difference in what the two methods offer, and no audit report fully closes it, because the audit is itself a procedure the voter must take on trust. The critics’ deepest point, that the system requires faith, cannot be refuted by describing the system more thoroughly. It can only be addressed by acknowledging that the faith is real and asking what would justify it.

What would justify it is the record of the system’s behavior under adversarial scrutiny, and here the defenders have their strongest material. The 2020 election was scrutinized more intensely, by more hostile observers, with more litigation and more recounts, than any election in American history, and the mail voting component survived. The North Carolina fraud was caught by the system’s own records. The rejection statistics are published. The processing is open to observers. A system that hides its failures cannot be trusted, but a system that publishes them, investigates them in public, and corrects them can be, not because publication proves perfection but because it proves the absence of concealment. This is the best available answer to the trust gap, and it is an institutional answer rather than a technical one: trust the system not because it cannot fail, but because its failures are visible and its corrections are public. Whether that answer satisfies the skeptics is not a question any procedure manual can settle. It is the question the next decade of election administration will have to live with.

What to Watch in Any Mail Ballot System

A reader who has followed the ballot’s journey, the federal rules, the history, and the evidence now has the tools to evaluate any mail voting system, and the evaluation can be organized around a short checklist of questions that cut through the slogans. First, which design is in force: universal, no-excuse, or excuse-required, because the design determines where the administrative burden falls and which evidence applies. Second, how does the jurisdiction verify the voter’s identity on the return envelope, and does it publish its rejection rates broken down by the categories the research has flagged. A system that verifies signatures but does not publish rejection data is asking for trust without offering the material trust is built on. Third, does the jurisdiction offer a cure process, how quickly does it notify voters, and what share of rejected ballots are ultimately counted. The cure statistics are the most revealing single number a mail voting system produces.

Fourth, when may the office begin processing ballots, and how long after Election Day does the count continue. A system that bars pre-processing until Election Day will count slowly, and the slowness is a legal choice rather than evidence of trouble, but it is a choice with trust consequences, because every day of counting is a day for suspicion to grow. Fifth, who may handle a voter’s ballot besides the voter, and what does the law require of those handlers. The collection rules are where the documented fraud has concentrated, and they are the detail most worth reading in any state’s statute. Sixth, what do the audits examine and what do they publish. A risk-limiting audit of the paper ballots is the gold standard, and its absence is not proof of failure, but its presence is the strongest single signal that the jurisdiction takes verification seriously.

The ballot’s journey table from the earlier section compresses the same checklist into the six stages, and it is worth keeping at hand whenever a new claim about mail voting circulates. Locate the claim in the table. Ask which verification is supposed to stop it. Ask what the published data show about that verification’s performance. The claims that survive that exercise are the ones worth worrying about, and they cluster, as shown above, in the return and verification stages, around collection practices and signature matching. The claims that do not survive it are usually claims about the counting machinery, which is the best-observed and least-failed part of the system. That distribution is the durable knowledge this subject offers. The machinery is knowable, its limits are measurable, and the remaining argument is about how much imperfection a democracy should accept in exchange for letting more of its citizens vote. That argument has no technical resolution, which is why it will outlast every procedure manual.

A companion reference is available.

Frequently Asked Questions

Q: What is the difference between absentee voting and vote-by-mail?

The two terms describe overlapping practices, and the distinction is mostly a matter of history and state law rather than mechanics. Absentee voting is the older term, dating to the Civil War practice of letting soldiers vote away from home, and it traditionally meant voting by mail for a stated reason such as illness, travel, or military service. Vote-by-mail is the broader modern term for any system in which the voter receives a paper ballot through the mail, marks it outside a polling place, and returns it for counting. In states with no-excuse absentee voting, any registered voter may request a mailed ballot, which makes the practice functionally identical to vote-by-mail even though the statute still says absentee. In the eight universal vote-by-mail states, every registered voter automatically receives a ballot, and the word absentee has largely fallen out of use. When a source uses one term or the other, the question to ask is which of the three system designs it describes: universal, no-excuse request, or excuse-required.

Q: Do all states let voters cast ballots by mail without an excuse?

No. States fall into three groups, and the excuse requirement survives in the smallest of them. Eight states mail a ballot to every registered voter automatically, with no request and no excuse needed. Most of the remaining states let any registered voter request a mailed ballot without stating a reason; this no-excuse absentee model is now the most common arrangement in the country. A minority of states still require the voter to give a qualifying reason, such as disability, illness, age over a threshold, or absence from the county on Election Day, before the election office will issue a mailed ballot. The exact membership of each group has shifted over time, with the general direction toward fewer excuse requirements, though a few states moved the other way after 2020 by tightening their rules. Because the categories change, the reliable way to learn the current rule is the voter’s own state election office, which publishes the request requirements and deadlines for each election.

Q: Why do some states require a witness signature on mail ballots?

A small number of states require the voter to have another adult sign the return envelope as a witness, and a few have required notarization, as an additional credential beyond the voter’s own signature. The stated purpose is to add a second human attestation that the ballot was completed by the voter, which addresses the concern that a mailed ballot is marked without supervision. In practice, researchers have found that witness requirements reduce participation, because voters must arrange for another person to be present at the moment of voting, which reintroduces a logistical hurdle that mail voting is meant to remove. Courts have weighed these requirements against the burden they impose, and several states dropped or suspended them after litigation. Most states concluded that the signature check, combined with the issuance log and the criminal penalties for fraud, provides sufficient verification without a witness. The witness rule survives where legislatures have judged the extra attestation worth its cost in convenience.

Q: What happens when a mail ballot signature does not match the record?

The envelope is set aside unopened, and the ballot inside is not counted unless the defect is resolved. Election workers compare the signature on the return envelope against the voter’s signatures on file, and when the comparison fails, the envelope goes into a rejected pile that is never opened, which preserves the distinction between a rejected voter and a rejected vote. In the more than two dozen states with a cure process, the office must notify the voter, explain the problem, and offer a cure affidavit, a sworn statement in which the voter attests that the ballot is theirs. The voter signs and returns the affidavit by a statutory deadline, usually a few days after Election Day, and the ballot is then accepted. In states without a cure requirement, the rejection is final, though the voter is typically notified. Because the envelope is never opened, a mismatched signature never affects the count; it affects only whether that voter’s ballot enters the count.

Q: Can a mailed ballot be tracked after it is returned?

In most jurisdictions, yes, through ballot tracking systems that function much like parcel tracking. Many election offices print an Intelligent Mail barcode on the return envelope, which the postal service scans as the envelope moves through the mail stream, and the office’s own intake scanning records the moment the envelope arrives and the moment its signature is accepted or rejected. Voters can usually check the status through the state or county election website or through a notification service, where available, that sends updates by text message or email. The tracking shows the ballot’s progress through stages such as mailed to voter, received by the office, signature verified, and counted. Tracking does not show how the voter marked the ballot, since the envelope is verified before it is opened and the ballot is separated from the voter’s identity afterward. Where tracking is unavailable, the voter can contact the election office directly to confirm receipt.

Q: Are mail ballots counted differently from in-person ballots?

No. Once a mailed ballot has been accepted and separated from its return envelope, it is indistinguishable from any other paper ballot and is counted on the same equipment under the same rules. Most jurisdictions use optical-scan tabulators that read the voter’s marks, and mailed ballots are fed through those machines either at a central facility or alongside in-person ballots. The tabulation equipment undergoes the same public logic and accuracy testing before the election, is observed by the same partisan observers during counting, and is subject to the same post-election audits afterward. What differs is everything before the count: the verification of the envelope, the separation of the ballot from the voter’s identity, and in many states the timing, since mailed ballots may be processed before Election Day or held until it. The counting step itself is the least mail-specific part of the pipeline, which is why disputes about mail voting rarely concern the tabulation.

Q: What is ballot harvesting, and where is it legal?

Ballot harvesting is the critics’ term for the organized collection of completed ballots by third parties, while election administrators usually call the same activity ballot collection. The practice sits on a spectrum. At one end, a family member dropping off a relative’s ballot is legal everywhere. At the other end, paid operatives collecting large numbers of ballots from strangers, as in the North Carolina case of 2018, is illegal in most states and was the conduct that led to that election being invalidated. Between those poles, state laws vary widely: some states let any person return another voter’s ballot, some limit collection to family or household members or designated agents, and some cap the number of ballots any one person may return. The variation reflects a genuine policy tension. Restrictive rules reduce the channel through which coercion and tampering can flow, but they also make voting harder for voters who cannot reach a mailbox or drop box themselves. Because the documented fraud has concentrated in this part of the pipeline, collection rules are among the most frequently litigated details of mail voting law.

Q: How common is documented fraud in mail voting?

Documented fraud is rare relative to the number of ballots cast, and that finding holds across the studies and databases that track it. The Heritage Foundation’s election fraud database, compiled by critics of mail voting, records on the order of a thousand proven cases across all voting methods over several decades, during which billions of ballots were cast. Academic researchers who have examined the same record describe the rate as a small fraction of one percent. The cases that do occur cluster in the return and collection stages rather than in the counting, and they typically involve small numbers of ballots, with the North Carolina congressional race of 2018 standing out as the exceptional case that affected an outcome. Defenders of mail voting argue that the low rate reflects both rare commission and effective detection, since the issuance logs and return patterns are designed to reveal anomalies. Critics argue that undetected fraud cannot be measured and that the published rate is therefore a floor. Both claims describe the same small numbers; they disagree about what the numbers leave out.

Q: Which states mail a ballot to every registered voter?

Eight states conduct all-mail elections in which every active registered voter automatically receives a ballot: Oregon, Washington, Colorado, Utah, Hawaii, California, Vermont, and Nevada. Oregon was the first, adopting the system by ballot measure in 1998, and the others followed over the next two decades, with the most recent adoptions becoming permanent in 2021. In these states, in-person voting has not disappeared; voters who prefer it can still cast a ballot at a voting center or drop off their mailed ballot in person. Several other states and many individual counties mail ballots to all voters for certain elections without having adopted the universal model permanently, and the list has changed over time as legislatures acted. Because the roster shifts, the eight-state count describes the durable core of universal systems rather than every jurisdiction that has ever mailed ballots broadly. The common feature of all eight is the automatic mailing; the request-based systems in the remaining states differ precisely on this point.

Q: Why do some mail ballot counts continue for days after Election Day?

The delay is a product of state law governing when processing may begin, not of anything going wrong with the ballots. Before a mailed ballot can be counted, its envelope must be verified, and states differ on when that pre-processing may start. Some states let offices verify and even scan ballots weeks before Election Day, holding the results until polls close, which produces a nearly complete count on election night. Others forbid any processing until Election Day itself, which means millions of envelopes must be verified, opened, and scanned while the in-person votes are also being counted, and the work necessarily stretches over days. States also differ on whether ballots postmarked by Election Day but arriving later may be counted, which extends the count by the length of the grace period. The variation is a policy choice about the tradeoff between speed and the safeguards of careful verification. Jurisdictions that moved their pre-processing dates earlier did so to shorten the count, and the ones that kept Election Day processing accept the delay as the price of their rules.

Q: What is a cure affidavit for a rejected mail ballot?

A cure affidavit is the sworn statement a voter signs to rescue a mail ballot that was rejected over a signature problem. When an election office sets aside a return envelope because the signature is missing or does not match the registration record, the cure process, required in more than two dozen states, obliges the office to notify the voter and offer this remedy. The affidavit typically asks the voter to attest, under penalty of perjury, that the ballot is theirs, and it may ask for a copy of an identification document. The voter signs and returns it by a statutory deadline, usually a few days after Election Day, and the office then accepts the ballot and counts it. The affidavit is attached to the original envelope and becomes part of the audit trail, so the rescue is itself documented and reviewable. The cure affidavit matters because it converts signature verification from a final judgment into a correctable step, and jurisdictions with well-run cure programs report that a large share of initially rejected ballots are ultimately counted.

Q: Are mail ballots rejected at higher rates for some groups of voters?

Yes, and the pattern is one of the best-documented findings in mail voting research. Studies of rejected mail ballots have consistently found higher rejection rates among young voters, first-time mail voters, and Black and Hispanic voters than among older white voters. Research on Florida’s elections found overall rejection rates near one and a half percent, with the burden falling disproportionately on voters whose signatures were least stable and who were least familiar with the cure process. The mechanisms are administrative rather than conspiratorial: young voters’ signatures change as they age, reference signatures on file may be years out of date, and voters with less experience of the system are less likely to know that a cure affidavit exists or that its deadline is short. None of this requires discriminatory intent by the workers applying the check. It requires only a verification method whose error rate correlates with voter characteristics. The finding is the main reason reform efforts have concentrated on the cure process, voter education, and alternatives to signature comparison.

Q: How did Oregon become the first all-mail voting state?

Oregon reached universal vote-by-mail through two decades of incremental experiment rather than a single decision. The state began conducting special elections by mail in the late 1980s, which gave administrators and voters experience with the mechanics on a small scale. Those experiments worked well enough that in 1998 voters approved Ballot Measure 60, making Oregon the first state to conduct general elections entirely by mail, and the measure passed by a wide margin with support across party lines. The state has run every election by mail since, which has produced the longest continuous record of all-mail administration in the country and the most mature version of the procedures other states later copied: centralized ballot tracking, signature verification with a cure process, drop boxes statewide, and public reporting of ballot status. Oregon’s experience functions as the natural experiment researchers cite most often, because no other jurisdiction offers a comparable span of data on fraud rates, turnout effects, and administrative costs under the universal model.

Q: Can a voter still vote in person after requesting a mail ballot?

Yes, in every state, because the system is designed with this fallback. A voter who requested a mail ballot but never returned it may vote in person, and the poll book check against the ballot issuance log confirms that no mail ballot was received before the in-person vote is allowed. A voter who returned the mail ballot and then attempts to vote in person will be stopped by the same check, since the log shows the ballot already received. A voter who spoiled or lost the mailed ballot can typically surrender it at the polling place, or sign an affidavit that it was not returned, and vote a regular ballot. These rules exist so that a lost, delayed, or defective mailed ballot does not disenfranchise a voter who can still reach a polling place. The issuance log does double duty here, guarding both the mail pipeline and the in-person rolls against the same ballot being counted twice, and it is checked at the moment of return and again at any in-person voting attempt.

Q: What is the Intelligent Mail barcode on a ballot return envelope?

The Intelligent Mail barcode is a postal barcode that many election offices print on mail ballot return envelopes so the ballot can be tracked through the mail stream like a parcel. The barcode encodes routing information that the postal service scans at each handling point, which lets the election office see when the envelope entered the mail and when it arrived, and lets the voter follow the same journey through the jurisdiction’s ballot tracking website. The barcode carries no information about how the voter marked the ballot; it identifies the envelope, not the vote. Its security function is indirect but real: because each barcode is unique to the issued ballot, it ties the physical envelope to the issuance log, which makes it harder to insert a ballot the office never sent. The barcode is one of several machine-readable features, alongside the voter’s signature and any required identification numbers, that turn the return envelope into a verifiable credential rather than an anonymous container.

Q: How do overseas military voters cast ballots by mail?

Members of the uniformed services, their eligible family members, and overseas civilians vote under the Uniformed and Overseas Citizens Absentee Voting Act of 1986, which requires every state to let them register and vote absentee in federal elections. The voter requests a ballot through the state’s process or the federal postcard application, and since the Military and Overseas Voter Empowerment Act of 2009, states must transmit the requested ballot at least 45 days before a federal election, a rule written after repeated findings that late-arriving ballots were disenfranchising service members. Many states now deliver the blank ballot electronically, though the voted ballot must still be returned by mail in most states. A voter who requested a ballot that never arrived may cast the federal write-in absentee ballot, a standardized backup ballot counted for federal offices. The overseas pipeline faces the longest transit times and the least reliable postal networks of any mail voting, and its documented problems have historically been logistical, involving late and lost ballots, rather than matters of fraud.

Q: Who is allowed to observe mail ballot processing?

Every state allows partisan observers to watch mail ballot processing, and the observation rights are among the most detailed in election law. Political parties and candidates may appoint observers who are entitled to be present when envelopes are received, when signatures are verified, when envelopes are opened, and when ballots are tabulated. Observers may challenge procedures and, in many states, individual ballot decisions, though they may not touch ballots or interfere with workers. Many jurisdictions go further, livestreaming the processing rooms or publishing the observation schedules so the public can watch without accreditation. The observation rules are the system’s answer to the invisibility objection: the voter cannot personally watch every step, but appointed representatives of competing parties can, and their presence is meant to ensure that no step happens on the authority of one side alone. Disputes about observer access, including how close observers may stand and what they may photograph, have produced recurring litigation, which itself indicates how seriously both sides take the watching.

Q: Why do election officials urge voters to return mail ballots early?

Early return protects the voter against the two most common reasons mail ballots go uncounted: late arrival and correctable defects discovered too late to fix. In states with receipt deadlines, a ballot that arrives after the polls close is not counted regardless of when it was mailed, and postal delays near Election Day make late mailing a gamble. Even in postmark states, a ballot mailed at the last minute may carry an illegible or missing postmark, which creates a dispute the voter cannot resolve from a distance. Early return also gives the cure process time to work: when an envelope is rejected for a signature problem days before the election, the office can notify the voter and the voter can return the cure affidavit before the deadline, but when the rejection happens on Election Day itself, the cure window may already be closing. Administrators therefore treat the published deadline as the last resort rather than the plan, and their guidance reflects the pipeline’s reality: the system works best when the ballot spends most of its life in the voter’s hands and the least possible time in transit.

Q: How do states prevent double voting by mail and in person?

The prevention rests on the ballot issuance log, the database in which the election office records every mailed ballot against the voter’s registration record. When a mailed ballot is returned, the office marks the voter as having voted, and any later attempt to vote in person is blocked when the poll book shows the mail ballot already received. When a voter appears in person without having returned the requested mail ballot, the poll worker confirms through the same log that no ballot is outstanding before issuing an in-person ballot. If both a mail ballot and an in-person vote somehow enter the system, the reconciliation during the official canvass, which compares the number of voters credited with voting against the number of ballots counted, flags the discrepancy for investigation. Attempting to vote twice is a crime in every state, and the log makes the attempt detectable rather than merely illegal. The system’s answer to the double-vote fear is therefore not trust but accounting: each voter gets one row in the log, and the row can only be satisfied once.

Q: What options exist when a requested mail ballot never arrives?

The system provides three remedies, in order of preference. First, the voter can contact the election office, which can check the issuance log, cancel the missing ballot so it can never be counted if it later surfaces, and issue a replacement. The cancellation is the key step, because it neutralizes the lost ballot while preserving the voter’s ability to vote. Second, the voter can vote in person, either during early voting or on Election Day, since the poll book check will show that no mail ballot was received and the voter remains eligible. Third, where the voter cannot reach a polling place, some states allow a ballot to be picked up in person by the voter or a designated agent close to the election. What the voter cannot do is simply vote twice and hope the system sorts it out; the issuance log treats the first ballot received as the vote and flags any second one. The existence of these fallbacks is why administrators describe a lost ballot as an inconvenience rather than a disenfranchisement, provided the voter acts before the deadlines.