Voting by mail in the US

The share of voters voting by mail jumped in the 2020 pandemic election, when many states eased their rules, and fell back in 2022 and 2024 (chart). The EAC's official counts show the same peak, from 43.1% in 2020 to 30.3% in 2024.

The EAC found the largest 2020 rises where states newly sent every voter a ballot or dropped excuse requirements.

In 2016 and 2018, mail ballots were "approximately one-quarter of voter turnout". In 2024 they were still "larger than the percentage of the electorate that voted by mail in pre-pandemic elections".

In 2024, early in-person voting reached its highest share in both the Census survey and the official counts.

In every election since 2018, voters aged 65 and over were the age group most likely to vote by mail (chart). The other age groups differ little from each other.

Most voters in the West vote by mail, and did so before the pandemic. In the South, mail stayed under a quarter of votes even in 2020 (chart). The Northeast and Midwest moved most in 2020 and fell back most after it.

Election offices' counts and voters' own reports agree on the shape: a peak in 2020, then a fall to about three in ten. Their levels differ by up to two and a half points, because one counts ballots and the other asks people.

NCSL expects seven more states to offer no-excuse or all-mail voting in November 2026 than in 2016 (chart). The count peaked in 2020, dropped by 2022 and has crept up since.

States have moved in both directions: several adopted all-mail voting after 2016, while Utah has voted to end automatic mail ballots (events under the chart).

Utah has sent every voter a ballot since all its counties adopted it in 2019, and in 2024 more than nine in ten of its votes came by mail.

A 2025 law changes that: from 2029 "a voter will not receive a ballot by mail unless the voter requests to receive a ballot by mail". A request lasts eight years. NCSL's table does not yet show the change.

Republican support for letting any voter vote by mail fell from about half in April 2020 to under a third in 2024 and 2025. Democratic support stayed above eight in ten (chart).

The first reading came early in the pandemic. Gallup links that year's fall in Republican confidence in elections to President Trump's "public questioning of the validity of mail voting". Neither survey tests whether his statements moved support for mail voting.

Oregon's statewide system came from a 1998 citizen initiative, after the state had tested mail elections locally since 1981. The other states on the list adopted it by bills passed by their legislatures.

Six of the nine first used it statewide from 2019 onward, around and after the pandemic; only Oregon, Washington and Colorado ran statewide mail elections before 2019.

Oregon, Washington, Hawaii, Utah and Colorado cast more than nine in ten ballots by mail in 2024. Vermont, DC and Nevada, which also mail every voter a ballot, cast between two-fifths and two-thirds that way (chart).

Arizona, which does not mail every voter a ballot, cast about three-quarters of its votes by mail.

Arkansas, Tennessee, West Virginia and Texas had the lowest mail shares of any state, each casting about one ballot in thirty or fewer by mail (chart).

In 2024 Democratic voters were about half as likely again as Republicans to vote by mail (chart). The gap has narrowed since 2020, "primarily due to Democrats pulling back in the use of voting by mail rather than Republicans increasing their usage".

In 2020 more than half of voters with disabilities voted by mail, a larger share than other voters (chart). Disabled voters who voted by mail were far less likely to report difficulty than those who voted at a polling place: 5% against 18%.

As California, Utah and Washington counties switched to universal mail ballots between 1996 and 2018, turnout rose by about 2 points of the citizen voting-age population (chart). That is a real but small effect, and it comes from years before mail voting split by party.

The same study found universal mail raised the Democratic share of turnout by only 0.1 points with trends included, with an upper bound of "only about 0.3%". Vote-share effects were "modest or null".

In Texas and Indiana in 2020, being old enough for no-excuse mail voting "did not substantially increase Democratic turnout relative to Republican turnout".

Texans who had just become old enough to vote by mail without an excuse turned out at almost the same rate as those a year younger (chart). They voted by mail more and early in person less.

In 2025, support for letting any voter vote by mail was highest in states that mail every voter a ballot and lowest in states that require an excuse (chart). The pattern holds within each party. The survey cannot say whether using mail voting builds support or supportive states adopt it.

Since 2012 the national rate has stayed between 0.8% and 1.5% of returned ballots. Even though twice as many mail ballots were sent out in 2020 as four years earlier, the EAC found no significant change from 2016 to 2020 or across the three elections from 2020 to 2024.

The same rejections look large as a count and small against all votes cast, since most voters did not vote by mail. Which denominator is quoted changes the impression, so both are given here.

A non-matching signature was the most common listed reason in four of the five elections from 2016 to 2024, and reached 40.7% of rejections in 2024. Late arrival led in 2012 and 2014 and again in 2018.

Envelopes with no signature were about a fifth of rejections in 2016 and about a tenth in 2022 and 2024.

Late arrival was the third-largest reason in 2024, behind signatures and "other". How many late ballots states count depends on their receipt deadlines, which a separate report covers.

Texas's line rises after SB 1 required an ID or Social Security number on the envelope, and has stayed above both its 2020 rate and the national line. In its first primary under the rule, many voters missed the new field on the envelope.

Georgia's line, high under signature matching in 2016 and 2018, has sat near the national line since SB 202. Those earlier rates count every reason, not just signatures. In a survey of county election offices, a plurality said the new check had not reduced rejections.

The two states moved in opposite directions under similar-sounding rules: Texas added a number check, Georgia used one to replace signature matching. So what a number check does to rejections depends on its design and on what it replaces.

In 2024, 11.1% of rejected mail ballots were from voters who had already cast another ballot that was accepted. The check stops a second ballot being counted; the survey does not say whether the second ballot was a mistake or an attempt to vote twice.

A missing witness signature was the next-largest specific reason, at 5.6%, and each remaining requirement accounted for less. None of the survey's reasons is fraud: every category is a failed procedural check.

Arkansas, Delaware and Oklahoma rejected the largest shares in 2024; Vermont, Maine and Indiana the smallest. Where a voter lives makes a large difference to the chance their ballot is not counted.

The EAC found the counted and rejected shares "not significantly different" between states that held all-mail elections and those that did not. The highlighted all-mail states sit throughout the middle and lower part of the chart.

Delaware checks only that the envelope is signed, Arkansas also requires a witness, and Oklahoma a notary; Texas checks an ID number as well as signatures. High rejection is therefore not a feature of signature matching alone. State rates by reason are not charted here.

Every group shown was rejected more often than white voters, and the gap was statistically significant for all but Native Americans. After controlling for other factors, the auditor estimated Black voters were twice as likely as white voters to have a ballot rejected.

These rates cover every rejection except late ballots, and almost all such rejections are for a mismatched or missing signature. The audit reports the age gap for mismatches alone, but not the racial gaps.

Men were 42 percent more likely than women to have a ballot rejected. In King County, voters casting non-English ballots were 47 percent more likely to be rejected than those casting English ones.

After controls, the youngest voters were almost three times as likely as the oldest to be rejected, and this held for mismatched signatures alone. First-time voters were rejected at more than five times the rate of voters who had voted before.

In Florida's 2018 election, younger voters and voters needing assistance were "disproportionately likely" to have mail ballots rejected, with "significant variation" across counties. Inexperienced mail voters were about three times as likely to be rejected in 2016 and 2018. Both come from the authors' summaries; the full papers were not read.

After controlling for age, race, education and income, ballots in some counties were four to seven times as likely as in others to be rejected for a mismatched or missing signature. For mismatches alone, two counties were 18 times as likely as another.

Counties with lower rejection rates were more likely to accept less conclusive signature matches, which could explain part of the difference. Some of the gap is a choice about how much doubt to allow, not only error.

Among the ten audited counties, those with lower rejection rates got 65 percent of flagged ballots cured, against 50 percent in the others.

In states that reported curing, more than half the ballots flagged for a fixable problem were corrected by voters and counted. Without curing, those ballots would have been rejected.

The EAC counted 43 states that let voters cure in 2024, but twelve of them could not report how many ballots entered curing, including Georgia, Oregon and Pennsylvania. The national figures cover only the states that reported.

In 2024 Texas sent 5.8% of its mail ballots into curing, against 1.5% nationally.

Most states compare signatures, but nine and DC check only that the envelope is signed, and six ask for an ID number. The groups overlap: Ohio and Texas compare signatures and also check a number. Utah will drop signature matching for ID digits in 2029.

Washington requires everyone who checks signatures to be trained on statewide standards, and Colorado has two more judges, of different parties, review any mismatch. All ten Washington counties the auditor visited had trained their staff.

Thirty-five states and DC require officials to tell voters of a missing signature or a discrepancy and give them a chance to correct it. The count includes states that do not compare signatures, which can only be curing missing ones.

Re-checking a random sample, the auditor agreed with counties on more than 98 percent of signatures. Most errors that remain are matters of degree: lower-rejection counties accepted less conclusive signatures.

In the auditor's words: "We found no evidence of bias in those decisions. While that is good news, it also means we are not able to explain what causes rejection rates to vary for these groups of voters."

The audit calls matching "inherently subjective". It found that "even experienced reviewers can come to different conclusions", that its own team disagreed on many signatures, and that counties read the state's criteria differently.

The court found that voters were not told their ballot had been rejected and moderators had no handwriting training or standards. The court noted that 275 of 78,430 absentee ballots there were rejected for a mismatch in 2016.

The majority held Texas's interest in stopping fraud "far outweighs" the burden, and quoted an earlier ruling that fraud risk is "much greater in the mail-in ballot context than with in-person voting". Its view that voting by mail carries no due-process interest was a prediction at the stay stage ("likely to show").

Judge Higginbotham, who concurred "only in the decision to stay", wrote that such "small" differences "have the potential to decide both local and national elections". The 5,000 figure is in his separate opinion, not the majority's.

The ruling interpreted the state's own election law, not the Constitution. Pennsylvania is now listed among states that do not compare signatures.

Kari Lake, who lost Arizona's 2022 governor's race by about 17,000 votes, asked the courts to install her or order a new election, claiming signatures had not been checked as the law required. The trial court found she had not proved it.

Georgia's Secretary of State and the state Bureau of Investigation re-examined 15,118 envelopes, about a tenth of the county's absentee ballots, and found "no fraudulent absentee ballots". Georgia used signature matching in 2020; SB 202 replaced it the next year.

The audit found two envelopes that should have been flagged. One was "mistakenly signed by the elector's spouse"; in both cases the voter had filled in the ballot. So a signature that was not the voter's got through, which shows a check can miss one.

The envelopes came from boxes of "accepted" ballots. The audit could find wrongly accepted ballots, but not valid ballots wrongly rejected, and finding no fraud in a sample shows little about how many forged signatures checks miss.

The sources gathered for this report contain no case of officials using signature review to change an election result. The one attempt found to overturn a result over signature checks, Arizona's 2022 governor's race, failed in court. The search was not exhaustive; see Data still needed.

In this report's hand count of Heritage's file, about two-thirds of the voided or overturned elections involved absentee or mail ballots, and half turned mainly on them. Heritage's list is a sample, so this is a share of known cases, not of all fraud.

Eleven were not. Georgia's 28th House District in 2018 was overturned over out-of-district voters, Compton's 2001 election over ballot order, and Pembroke, North Carolina, redid elections over residency.

None of the 32 is in Colorado, Oregon, Utah or Washington. Their entries are individuals, often casting a dead or absent relative's ballot; one Washington man was caught because his wife's signature did not match.

Most were city, county or primary races decided by a few dozen votes or fewer; Caddo Parish's sheriff runoff was decided by one. The one federal race, North Carolina's 9th District, was decided by 905 votes.

In each charted case the invalid ballots outnumbered the margin, except East Chicago. There the Indiana Supreme Court ordered a new election anyway, holding that the scheme "perverted the absentee voting process", whether 155 or 100 ballots were invalid.

Two separately built compilations, Heritage's and News21's, each find absentee-ballot fraud many times more often than voting in person under another's name.

Heritage's absentee entries span four decades. In the 2024 general election alone, 47,629,437 returned mail ballots were counted. No source turns the case counts into a rate per ballot.

North Carolina's elections board found its 2018 congressional race "corrupted by fraud, improprieties, and irregularities so pervasive" that it ordered a new election. Courts did the same in East Chicago, Bridgeport and elsewhere (see the table).

Every scheme in the table whose discovery is recorded was exposed by an election contest, an investigation or a paper trail, and the remedy was a new election. No statewide or presidential result in this record was overturned for absentee fraud.

A judge ordered Paterson's May 2020 council election rerun. The winner's indictment was dismissed in April 2026 on speedy-trial and disclosure grounds; the state is appealing. As of May 2026 no one has been convicted.

The 2005 bipartisan commission wrote that "absentee ballots remain the largest source of potential voter fraud", and that voters at home or in nursing homes are more open to pressure. It urged states to stop candidates and party workers handling absentee ballots.

It found "little evidence of fraud in Oregon, where the entire state votes by mail", and "no evidence of extensive fraud in U.S. elections". It still asked Oregon for better precautions against intercepted ballots.

In Lincoln County, West Virginia, the county clerk himself falsified absentee ballots and stuffed boxes, and in Philadelphia city commissioners were found complicit. Chicago's 1982 fraud was done by election judges.

In Otero and Cochise counties in 2022, officials who refused or delayed certification were ordered to certify, and did. Two Cochise supervisors were indicted. Refusals since 2023 were not searched.

The record read contains no case of officials rejecting ballots on signatures to change a winner. The closest is the reverse: Arizona's losing candidate for governor claimed in 2022 that signatures were not checked, and the court found she had not proved it. The search was incomplete (see Data still needed).

Kansas City's 1936 fraud and Chicago's 1982 fraud, both run by officials at the polls, each produced more convictions than any absentee case in this report. Absentee fraud is the larger category in modern compilations, but not the only large-scale kind on record.

Germany's postal share rose from about one voter in twenty in 1957 to more than a quarter in 2017, and jumped to almost half in 2021. It fell back in 2025; the source gives no reasons for either move.

Since 2008 German voters have not had to give a reason. The Constitutional Court noted the legislator's view that a significant rise in postal voting "was not to be feared" from the change; the share later rose sharply.

In 2013 the court held that with postal voting "public control of how the vote is cast is reduced" and integrity "is not guaranteed to the same degree" as at a polling station.

It upheld it as serving the widest possible participation, adding that lawmakers must ensure no election principle is "substantially hollowed out", which it found "plainly not the case at present".

Cantons report that "around 90%" of Swiss citizens vote by post, which has not led to "major political debates about security or fraud".

Cases occur: in Bern 300 votes were voided in 2016 for identical handwriting, and an Upper Valais man was convicted in 2018 of forging at least 193 ballots.

For half a century postal votes were a few percent of the UK vote. After voters no longer had to give a reason, in 2001, they reached about a fifth of valid votes by 2010 and stayed there.

The Tower Hamlets election court wrote that "postal voting on demand made personation once again viable". The Birmingham judgment of 2005 had found "literally thousands of votes" falsified, perhaps half of the winners' votes in two wards.

Only 11 of the 204 countries International IDEA codes let all voters vote by post at home.

Another 21 let some voters. Among OECD members it is more common: 27% of them offer it to all voters.

The 11 include the US, the UK, Germany, Switzerland and Canada. France and Australia allow it only for some voters.

IDEA notes that letting voters correct an invalid ballot, as many US states do, "is uncommon globally".

The Interior Ministry said in 1988 that every party had condemned postal voting as the occasion of "fraudes graves" (serious fraud) after each election. With no voter ever appearing in person, it allowed votes for "electeurs fictifs" (fictitious voters; the source prints French without accents). IDEA gives the same reason.

The 1975 law abolished postal voting while widening voting by proxy at the same time, so remote voting did not simply end. IDEA's 2025 list counts France among countries allowing postal voting for some voters.

Abstention fell at presidential, municipal and cantonal elections and rose at legislative ones, which the ministry put down to low turnout in June 1988. These are a government's own averages, not a test of cause.

The National Assembly rejected a bill to restore postal voting on 3 February 2022. A 2023 Senate bill to allow it for presidential elections has lapsed.

Before it, "the vote buyer could simply place a ballot in the hands of the bribed voter and watch until he placed it in the polling box". Reformers argued secrecy would end bribery's payoff and protect "the weak and dependent against intimidation and coercion by employers and creditors".

Before Chile's 1958 reform, places with more patron-client ties voted more strongly for the right; afterwards "this difference across localities completely disappeared". The abstract gives no figures.

These reforms secured secrecy inside a polling booth. No study read here measures vote buying or coercion among mail voters, so whether mail voting revives them is not shown either way.

In the strongest survey, over 23 percent of mail voters had someone else in the room and more than six percent showed their ballot. Mail voters were likelier than Election Day voters to doubt their vote was secret.

Letters assuring people their ballot was secret raised turnout by an estimated 3.4 percentage points among people who had never voted. People who doubt secrecy are more swayed by social pressure in how they vote.

Its questions count lawful help to a spouse or child as filling out a ballot, and helping a disabled voter is "perfectly legal in every state". Its opt-in sample, small mail subsample and low mail share make even its secrecy figures far weaker than the 2010 survey's.

In the Oregon reports "1%, or fewer" felt pressured by another person's presence. Oregon researcher Priscilla Southwell wrote that there had been "no evidence that undue influence on voters – like bribes or threats – has been a problem".

Britain's 2016 fraud review heard that household undue influence goes unreported for reasons like those of domestic-abuse victims. The Oregon pilot's author suspected her figures were too low.

In the same 2010 survey, 22 percent of in-person voters had no privacy screen. UK observers saw family voting at 21% of polling stations in 2024.

In Birmingham in 2005 a judge found "literally thousands of votes" falsified, calling the problem "universal". Election courts voided a Woking ward election in 2013 and Tower Hamlets' 2014 mayoral election. Slough's 2008 case was register fraud rather than postal fraud.

It responded by requiring a signature and date of birth (2006), banning campaigners from handling postal votes and making postal votes lapse (2022). Northern Ireland does not offer postal voting on demand.

The Electoral Commission found "no evidence of large-scale electoral fraud" in 2019–2023: 11 convictions from 1,462 allegations of every kind. Like all detected counts, these are a floor (see the US fraud report).

In 2024, 2.2% of returned postal votes were rejected because the signature and date of birth could not be verified, a level "broadly similar" to earlier polls. Britain's check pairs a signature with a fact that needs no judgement, the date of birth.

Of the 3% of postal voters who had help, 9% gave that reason. The Commission said it could not judge how far this was unwanted, but "some of these instances" may have been coercive.

Some ballots arrived late, but it found "no widespread, systemic issues with postal voting", and 89% of postal voters were satisfied.

Germany's court upheld postal voting while conceding reduced control and integrity. Britain's 2016 review recorded the argument that unrestricted postal voting "can never guarantee that the voter has not been influenced". Where they differ is whether the loss has, in practice, led to coercion.

Observers who count family voting at polling stations say they "are unable to collect data" on it for postal votes. The surveys above measure presence and help, not pressure (see Data still needed).

The number of states counting late-arriving ballots peaked in November 2020 and has fallen since (chart). NCSL expects the count of states requiring Election Day receipt in 2026 to be "roughly the same as it was in November 2016".

North Dakota, Ohio and Utah moved to Election Day receipt in 2025. Kansas legislated the same change, but a court has blocked it for now (both marked under the chart).

The states still counting postmarked ballots that arrive late include Texas, Alaska, Kansas, Mississippi and West Virginia, as well as California, New York and Illinois. Windows run from one day in Texas to about three weeks in Washington (next chart).

In every federal general election since 2016, between about one in eight and more than one in four rejected mail ballots was rejected for arriving too late (chart).

In 2024 that was 17.8% of rejections, the third-largest reason after signature problems and "other". Rejections themselves were 1.2% of returned mail ballots.

Late arrival's share of rejections was lowest in 2020, when the most states counted ballots arriving after Election Day (charts). The two series come from different sources, and much else about voting changed in 2020, so the timing alone does not show that grace periods caused the drop.

The Supreme Court described Mississippi as "one of roughly 30 States" counting late-arriving ballots. NCSL's 15 states are those that do so for ordinary domestic mail ballots.

Its separate count for military and overseas voters is "Thirty states and Washington, D.C.", which accept at least some such ballots after Election Day. Both numbers are right; they count different ballots.

In 2024 a majority of the national vote was counted in about two hours and three-quarters in about three. Every state had reported more than 75% of its presidential vote within about five days.

MIT names Washington, California, Nevada, Arizona, Utah, Colorado and Alaska as having the most left to count after a day. California reported "about a third of its votes" over the month after Election Day.

Texas and Florida, each with well over ten million votes, counted almost all of them within about a day.

Mississippi, Maine, Maryland, Massachusetts and Michigan were also slow to report in 2024. Among the battleground states, Nevada and Pennsylvania counted far faster in 2024 than in 2020, while Arizona took twice as long (chart).

MIT's 2020 study found counts slower where states had "large volumes of mail ballots, prohibited preprocessing mail ballots, and allowed mail ballots to arrive after Election Day".

The Electoral Integrity Project's expert survey of the 2024 election rated "the counting, results and procedural aspects" strong, while overall US scores fell. It measures experts' judgement of the process, not public confidence (see confidence and disputed results).

In 2020, states that let officials prepare mail ballots before Election Day had reported more of their vote four hours after polls closed (chart). The number of states prohibiting this fell from seventeen in 2020 to seven in 2024.

By 24 hours after polls closed, "the gap between states had essentially disappeared". Arizona and Nevada process mail ballots on receipt and were still among the slowest states in 2024 (table below).

Each row uses a different election and denominator, so the figures are not comparable with one another.

Less than 1 percent of ballots cast in Nevada's two largest counties arrived after Election Day, and Senator Rosen's margins there "far" exceeded them.

Against mail ballots rather than all ballots the share is larger: the two counties' late ballots alone come to nearly one in fifty of the mail ballots returned statewide.

In Wisconsin, Michigan and Washington, more mail ballots arrived too late in 2026 primaries than in the previous comparable election (table). None of these ballots was counted.

Washington, an all-mail state with a grace period, shows that late rejections happen under postmark rules as well as under Election Day deadlines.

The Postal Service says more than 95% of mail ballots have arrived on time in recent elections. Since 2024 it postmarks mail at regional centres rather than where it is posted, which the reporting says has "particularly affected" states that rely on postmarks.

Senator Rosen trailed on election night in 2024 and took the lead the next night, once about 50,000 Clark County mail ballots received on Election Day were counted. The late swing came from the time taken to count, not from ballots arriving after Election Day.

Processing rules are NCSL's as of September 2026 and may differ from those in force in 2020; mail share is from EAVS 2024.

Most states let officials begin processing mail ballots before Election Day; six, including Pennsylvania and Wisconsin, allow it only on Election Day. No state allows results to be released before polls close.

In Watson v. RNC, decided by five votes to four, the Court held that federal election-day statutes do not stop states counting ballots that arrive afterward (marked on the first chart).

On the arguments that Election Day receipt protects integrity and confidence, it said such "policy arguments are properly directed to legislatures, not courts".

The opinion said legislatures may have set Election Day deadlines because "an election-day deadline avoids the appearance of fraud from late-arriving ballots". It did not decide whether that view is right: "The question today is not whether requiring ballots to be received by election day is a good or bad idea".

Republicans were more confident than Democrats under George W. Bush, and Democrats more confident under Barack Obama and Joe Biden (changes of president are marked under the chart). The gap reached a record 56 points in 2024.

A study pooling about 30 national surveys from 2000 to 2012 finds "a consistent relationship between voting for the winner and confidence in election administration".

Gallup's 2020 reading was taken before the election. It ties that year's fall in Republican confidence, under a Republican president, to "then-President Donald Trump's public questioning of the validity of mail voting".

After the result, Republican confidence fell further, to 22% in MIT's post-election survey (chart below).

More than half of Americans have been confident in every Gallup reading since 2004. But the share "not at all confident" rose from 6% in 2004 to 19% in 2024.

In each of Pew's four October readings since 2020, fewer voters were confident that mail and absentee votes would be counted as intended than votes cast in person (chart, both panels).

At most 42% of Republican candidates' supporters have been confident in the mail count in any reading.

Democratic confidence in it fell from 85% among Harris voters in 2024 to 64% in October 2026.

In-person confidence also rose among Republicans and fell among Democrats between 2024 and 2026.

About three in four mail and absentee voters are confident their kind of ballot will be counted as intended, against about half of in-person voters (chart). Confidence in the mail count is highest among the people who use it.

In MIT's post-election survey, each party's confidence in the nationwide count rises when its candidate wins the presidency and falls when he or she loses (chart).

In surveys from the 2000s, winners' confidence in their own vote rose about 20 points after an election, against a fall of about 2 points among losers.

A study of 2016 finds the winner effect outweighed pre-election "rigged" messaging from elites. In 2018 the usual winner effect did not appear, and in 2008 both parties' confidence fell to 57%.

Republican confidence was already low before the 2020 count and fell further after it. Because the same voters also lost, these series cannot show whether a long count or a changing lead lowers confidence on its own; see Data still needed.

US confidence in the honesty of elections has averaged 45% since 2006, with its low in 2016 (chart). This is a different question from Gallup's vote-count item and gives lower figures.

Party was recorded only from 2024. The Democratic–Republican gap was 55 points that year, closed to 4 points in 2025 and reopened in 2026.

Across more than 100 countries, people who approve of their national leadership are far more confident in election honesty than those who disapprove (chart).

In 2026 American disapprovers were the more confident group, which Gallup says no other country has shown in 15 years.

These are prominent recent cases in which a losing side challenged a result, not a complete list.

The 2020 dispute followed months of questioning mail voting, Arizona's centred on signature checks of mail ballots, and Bridgeport's on absentee ballots in drop boxes (table). The three earlier disputes were about a recount, irregularities at the polls and the voter rolls.

Five disputes were settled by a court or by Congress; in Georgia the losing candidate ended her campaign without conceding. In every case the original winner ended up in office. In Arizona, the losing candidate "did not contest whether voters' signatures on ballot envelopes matched".

Unlike the others, Bridgeport's challenge succeeded: the judge found the miscounted ballots left the court "unable to determine the result of the primary". The incumbent won the court-ordered rerun primary and the later general election.

The Gallup, Pew and MIT series above are annual, pre-election or post-election snapshots, and Florida 2000 predates them. None measures confidence immediately before and after a dispute, so whether disputes themselves lower trust cannot be read from them.

No study found tests whether tighter mail rules raise the losing side's confidence in results. The nearest evidence points both ways at once.

Clinton finds voters "less likely to express confidence in election results conducted using procedures they oppose", beyond who wins. In Georgia, 59.1% of Republicans said the state's 2021 election law, which tightened ID rules for mail ballots, raised their confidence, while a 46.8% plurality of Democrats said it lowered theirs.

Georgia Republicans' confidence in the statewide count rebounded in 2022, the first general election under that law (SB 202). The authors caution this may be "the noted winner's effect", since the party won almost every statewide office that year.

The number of states offering mail voting to every voter rose sharply for 2020, fell back as temporary rules ended, and is expected to be seven above its 2016 level in November 2026 (chart).

Within that rise, states pulled the other way too. Utah, which mails every voter a ballot, will send ballots only on request from 2029. Oregon's all-mail system was adopted by its own voters, by citizen initiative in 1998.

More states counted late-arriving mailed ballots in 2020; by November 2026 NCSL expects the count of states requiring Election Day receipt to be "roughly the same as it was in November 2016" (chart).

North Dakota, Ohio and Utah ended their grace periods in 2025. Kansas legislated an end too, but a district court in Kansas blocked it in July 2026, so Kansas still counts ballots arriving within three days. Detail by state is in the report on ballot deadlines.

The Democrats' Freedom to Vote Act, which would have required every state to offer no-excuse absentee voting, failed to end debate in October 2021 and again in January 2022.

In August 2026 a Republican photo-ID bill, S. 5271, also failed to reach 60 votes (chart). Bills that would require or restrict mail voting nationally have been introduced on both sides in the current Congress.

UOCAVA requires every state to let service members, their families and Americans abroad "register and vote absentee in federal elections". Since a 2009 amendment, the year of the MOVE Act, a state must send a requested ballot "not later than 45 days before the election".

These are national mail-voting rules, but for a defined group. For everyone else, Congress has left the choice of no-excuse, excuse-only or all-mail voting to the states.

The House passed the SAVE America Act, 218–213, in February 2026 and again in July with the defence bill. An advocacy outlet says the July bill "would end universal mail voting in eight states and DC".

The Congressional Research Service, writing on 10 July, listed a ban on sending ballots except on request only as "a proposed amendment to the House-passed" bill. The bill text was not read.

The Universal Right to Vote by Mail Act and the Vote at Home Act would require no-excuse mail voting and all-mail elections. The MEGA Act and the Mail Ballot Integrity Act would ban sending ballots except on request and limit mail voting to specified voters. None has passed.

Republicans looked to budget reconciliation, which cannot be filibustered, as a route for the bill; a budget plan was paused in August when President Trump intervened. The photo-ID bill rejected on 8 August was a separate measure, S. 5271.

The Elections Clause gives the "manner of holding elections" to each state's legislature, subject to Congress's power to "make or alter such regulations". The President's part is to sign or veto bills.

Moore v. Harper added that state courts can review a legislature's election rules, so "the states" means each state's whole constitutional system.

The Supreme Court has held that Congress's power over the times, places and manner of congressional elections "is paramount, and may be exercised at any time, and to any extent which it deems expedient".

The same ruling drew a line at "who may vote", which stays with the states. Who may vote by mail, as opposed to who may vote, has not been tested against that line in the sources read.

The Court held, 5–4, that "nothing in the federal election-day statutes requires ballots to be received by election day". It said policy arguments "are properly directed to legislatures, not courts".

It pointed to Congress: "If varied deadlines for ballot receipt similarly call for a national solution, the American people must choose it through their elected representatives." It did "not consider the scope of Congress's authority".

The Court noted that legislatures "may have shared plaintiffs' view that an election-day deadline avoids the appearance of fraud from late-arriving ballots". It also read UOCAVA as confirming that "States have the power to set the date by which they must be received".

A federal court in Massachusetts blocked the Postal Service's final rule on 4 September, and the Supreme Court refused to lift that order on 14 September. The Court said the government "is unlikely to succeed on the merits of its challenge".

Justice Kavanaugh, concurring, gave a narrower reason: the rule may be within the Postal Service's authority, but election officials lacked "sufficient time to reasonably implement the rule before the elections". The rule could return for later elections.

Justices Alito and Thomas dissented: "Even assuming the rule here is a regulation of elections, Congress may create such rules under the Elections Clause. And, by statute, it may authorize the Postal Service to issue such rules." In their view the dispute was about the Postal Service's statutory powers, not the states' constitutional ones.

In August the Supreme Court lifted an injunction against Executive Order 14399, calling it "an internal directive" that "neither requires nor forbids anything of anyone outside the Executive Branch". It added that this "does not mean that any measure taken by the Government to implement the Order will necessarily be lawful".

The dissent said the ruling "merely postpones adjudication" of whether the President's actions are lawful.

The federal court in Massachusetts held that the challenged sections of Executive Order 14248 "exceeded the President's constitutional authority" and conflicted with federal statutes. Its July 2026 injunction covers the order's "Election Day ballot receipt deadlines"; the government has appealed.