Voter Fraud Statistics: Types, Rates, and Policy Impact

Voter fraud statistics from decades of audits, academic studies, and court records point to the same conclusion: proven fraud in United States elections is extraordinarily rare, with documented rates ranging from roughly 0.0003% down to 0.00000013% of ballots cast, depending on the study and the type of fraud examined.¹ The Brookings Institution has concluded that no U.S. election outcome has ever been altered by ballot fraud.²

How Often Fraud Actually Happens

The Brennan Center for Justice reviewed carefully studied elections and found fraud incident rates between 0.0003% and 0.0025%. A 2014 analysis published in the Washington Post identified 31 credible instances of voter impersonation out of more than one billion ballots cast between 2000 and 2014.¹

Brookings, working from the Heritage Foundation’s own election fraud database, ran the math at the state level. Arizona recorded 36 fraud cases over 25 years against more than 42.6 million ballots, a rate of 0.0000845%. Pennsylvania produced 39 cases across 30 years and more than 100 million votes.²

A Department of Justice investigation running from 2002 to 2004 produced a fraud rate of 0.00000013% of ballots cast and found no evidence of in-person impersonation.¹ Looking at a longer window, from fiscal year 2001 through 2017, the DOJ’s Public Integrity Section opened 33 election fraud matters and filed 19 cases, around 3% of its workload. U.S. Attorneys’ offices opened 525 matters and filed 185 cases, about 0.02% of their caseload.³

What Kinds of Fraud Show Up Most

The categories are not equal. The News21 investigative project analyzed 2,068 alleged election fraud cases from 2000 to 2012, drawing on more than 2,000 public records requests and nearly 5,000 documents. The breakdown:⁴

Absentee ballot fraud accounted for 491 cases, about 24% of accusations and the single largest category.
Registration fraud accounted for 400 cases, about 19%.
Ineligible voters, meaning felons and noncitizens combined, accounted for 393 cases.
In-person voter impersonation accounted for 10 cases over the full 12-year period, less than 1%.

Pew, drawing on the same News21 dataset, put voter impersonation at just 0.5% of cases studied.⁵ The Brennan Center has called impersonation “the rarest fraud of all,” observing that an American is more likely to be struck by lightning than to impersonate another voter at the polls. Many incidents that look like fraud on first inspection turn out to involve clerical errors, poor data matching, or administrative mistakes.¹

Of the News21 cases where resolution could be determined, 46% ended in acquittals, dropped charges, or decisions not to prosecute.⁴

Mail and Absentee Ballots

Mail voting has drawn the most sustained scrutiny since 2020. A Brookings study published in November 2025 looked at fraud cases from the 2016, 2018, 2020, and 2022 general elections and calculated an average mail voting fraud rate of 0.000043%, roughly four cases for every 10 million mail ballots cast. Each cycle produced between six and 46 identified cases.⁶

Universal vote-by-mail systems, where every registered voter automatically receives a ballot, produced the smallest share of mail voting fraud compared with traditional absentee systems.⁶ A 2023 white paper from the MIT Election Data and Science Lab reached the same conclusion: fraud in state mail elections is negligible.⁷ In 2024, 29% of voters used mail ballots, and the Postal Service processed more than 99.2 million of them.

One point election administrators emphasize: ballot rejection is not fraud. Ballots are rejected for missing signatures, mismatched signatures, or late arrival. Those are administrative outcomes, not evidence of wrongdoing.⁷

Noncitizen Voting

Noncitizen voting is a frequent subject of fraud claims and a frequent subject of state audits. It has been illegal for noncitizens to vote in federal elections since 1996, with penalties of up to five years in prison and potential deportation.⁸

The Heritage Foundation database identified 23 instances of noncitizen voting from 2003 to 2022.⁸ The Cato Institute, citing the same source, counted roughly 85 cases since 2002.⁹ A Brennan Center survey of 42 jurisdictions covering 23.5 million votes in 2016 found about 30 suspected noncitizen incidents referred for investigation, roughly 0.0001% of votes cast.¹⁰

State audits tell a similar story. Georgia’s 2022 citizenship audit flagged 1,634 noncitizens who had attempted to register, and the state confirmed none had cast a ballot.¹¹ Utah reviewed around 2.1 million registered voters and found one confirmed noncitizen, who had never voted. Idaho identified 36 likely noncitizen registrants among more than a million voters, which its Secretary of State described as “10 thousandths of a percent.”¹² Louisiana found 390 noncitizen registrants out of 2.9 million, with 79 having voted at least once over several decades; the state’s secretary of state concluded noncitizen voting is “not a systemic problem.”¹³

Large-scale screening tools have their own problem: false positives. The Department of Homeland Security recently processed 49.5 million voter files through the SAVE program and flagged roughly 10,000 registrants as potential noncitizens. Experts place expected error rates for data matching at scale between 0.9% and 2.5%, which means many of those flagged are likely citizens. U.S. Citizenship and Immigration Services acknowledged providing incorrect information to at least five states. In Missouri, more than half of the voters flagged were confirmed to be U.S. citizens.¹³ A Texas investigation concluded that more than 5% of voters identified by SAVE as potential noncitizens were actually citizens.¹⁴

The Heritage Database and Its Critics

The most frequently cited catalog of proven fraud is the Heritage Foundation’s election fraud database, first published in 2017 and described by Heritage as a “sampling of proven instances” rather than a comprehensive record. Its most recent update, in December 2025, listed close to 1,600 documented cases spanning decades and all 50 states.¹⁵ The database catalogs fraudulent use of absentee ballots, false registration, duplicate voting, and other categories.¹⁶

The Brennan Center assessed the database in 2017, when it held roughly 1,100 entries, and called the claims drawn from it “grossly exaggerated and devoid of context,” noting many entries were only tangentially related to voter fraud and that the total amounted to a “molecular fraction” of votes cast.¹⁷ Brookings reached a similar conclusion, saying that measured against the hundreds of millions of votes cast over the same decades, the numbers are “minuscule.”²

What Investigations of the 2020 Election Found

The 2020 presidential election produced the largest body of fraud allegations in recent memory, and also the most thorough testing of them. The Trump campaign paid software engineer Ken Block approximately $750,000 to analyze data from Arizona, Georgia, Nevada, Michigan, Wisconsin, and Pennsylvania. Block found no evidence of mass voter fraud. Many claims, he concluded, rested on misinterpreted or incomplete data, such as mail-in ballots incorrectly flagged or voters with similar names counted as duplicates. Across all six states combined, Block identified fewer than 200 duplicate mail-in ballot votes that were fraudulently cast.¹⁸ Block later wrote that the campaign “did not want to hear” his findings.¹⁹

A peer-reviewed study published in the Proceedings of the National Academy of Sciences in November 2021 evaluated the statistical arguments made by Trump and his supporters, examining claims about county-level vote distributions, bellwether counties, Dominion voting machines, absentee ballot irregularities, and turnout anomalies. The researchers found none of the statistical claims convincing. In every case, the supposedly anomalous patterns were either factually wrong or fully explained by normal democratic trends, including the well-established concentration of Democratic votes in urban areas.²⁰

An earlier federal effort reached the same empty result. The Presidential Advisory Commission on Election Integrity, established in May 2017 and chaired by Vice President Mike Pence with Kansas Secretary of State Kris Kobach as vice chair, held two meetings before Trump dissolved it in January 2018.²¹ Maine Secretary of State Matthew Dunlap, a commission member, sued for internal documents, reviewed more than 8,000 pages, and concluded the commission “uncovered no evidence of widespread voter fraud.” A draft of the final report included an outline for a section intended to present fraud evidence, but the section contained no actual content.²²

Prosecutions and Penalties

Individual prosecutions continue. The Heritage Foundation database lists criminal convictions for 2025 in Florida, Missouri, Alaska, Alabama, California, Colorado, West Virginia, New Jersey, and Pennsylvania.²³ In May 2025, federal prosecutors in Massachusetts indicted Lina Maria Orovio-Hernandez, a Colombian national, on charges of fraudulent voter registration, fraudulent voting in the November 2024 election, identity theft, and stealing over $400,000 in federal benefits.²⁴ Heritage tracked 43 instances of alleged or proven voter fraud across the 2024 election cycle.²⁵

Under 52 U.S.C. § 20511, knowingly and willfully submitting voter registration applications or casting ballots known to be materially false, fictitious, or fraudulent carries a maximum penalty of five years in prison and a fine.²⁶ The FBI groups federal election crimes into fraud by voters (false registration, voting while ineligible, double voting, impersonation), fraud by officials, bribery, intimidation, and campaign finance offenses.²⁷

How the Statistics Drive Policy

The gap between recorded fraud and the laws built around it is where this debate actually lives. In 2008, the Supreme Court upheld Indiana’s voter ID law in Crawford v. Marion County Election Board. The Court acknowledged there was no evidence that in-person impersonation fraud had ever occurred in Indiana, but held that states have a legitimate interest in deterring fraud and maintaining public confidence, and that ID requirements impose only a “limited burden” on most voters.²⁸

Research from the MIT Election Data and Science Lab has found that strict photo ID laws correlate negatively with turnout and disproportionately affect racial minorities, who are more likely to lack qualifying identification. After 2020, Georgia, Florida, and Montana increased ID requirements for mail-in voting.²⁹

At the federal level, the Safeguard American Voter Eligibility (SAVE) America Act would require documentary proof of citizenship for voter registration and strict photo ID for federal elections. The House passed the measure in February 2026, and the Senate began debating a companion version in March 2026. The bill would also require states to use the SAVE verification system to screen voter rolls and would impose criminal penalties on election officials who register applicants without proper documentation. It contains no federal funding and no phase-in period.³⁰

What Voters Believe Versus What the Data Shows

Perception does not always track evidence. The 2024 Survey of the Performance of American Elections, conducted by the MIT Election Data and Science Lab with 10,200 registered voters, found that respondents generally believe voter fraud is rare, though attitudes remain polarized along partisan lines. The same survey found 96% of voters were confident their own vote was counted correctly, and the partisan gap in confidence narrowed in 2024.³¹³²

The Cato Institute, writing from a libertarian perspective, has reached the same factual conclusion as the Brennan Center: noncitizens do not vote illegally in detectable numbers, and claims to the contrary are not supported by data from any state that has conducted a rigorous audit.¹² On the numbers, researchers across the political spectrum describe the same picture.

1
Brennan Center for Justice. Debunking the Voter Fraud Myth
2
Brookings Institution. How Widespread Is Election Fraud in the United States? Not Very
3
U.S. Government Accountability Office. Elections: DOJ and DHS Stakeholder Coordination (GAO-19-485)
4
News21. Election Fraud in America
5
Pew Charitable Trusts. Election Fraud
6
Brookings Institution. Mail Voting in the US: Data Points to Very Low Fraud and Significant Benefits to Voters
7
SciLine. Voting by Mail
8
Migration Policy Institute. Noncitizen Voting in US Elections
9
Cato Institute. The Rights and Bogus Claims About Noncitizen Voting Fraud
10
Brennan Center for Justice. Noncitizen Voting Is Vanishingly Rare
11
Georgia Secretary of State. Citizenship Audit Finds 1,634 Noncitizens Attempted to Register to Vote
12
Cato Institute. Trump’s Claims About Noncitizens Voting Are False. We Can Prove It
13
Brennan Center for Justice. Watch Out for False Voter Fraud Claims Fueled by the SAVE Program
14
FactCheck.org. Flaws in Government Tool to ID Noncitizen Voters
15
Heritage Foundation. 2024 Annual Report
16
Heritage Foundation. Election Fraud Cases
17
Brennan Center for Justice. Heritage Fraud Database: An Assessment
18
Business Insider. Ken Block Hired by Trump Campaign to Investigate Election Fraud
19
KUNR. Disproven: Trump Hired Ken Block to Investigate Election Fraud in 2020. He Didn’t Find Anything
20
Proceedings of the National Academy of Sciences. No Evidence for Systematic Voter Fraud: A Guide to Statistical Claims About the 2020 Election
21
Brennan Center for Justice. Disbanded: Trump’s Voter Fraud Commission
22
American Oversight. Voter Fraud Commission Records Show
23
Heritage Foundation. Election Fraud Cases Search
24
U.S. Department of Justice. Colombian National Charged With Voter Fraud, Federal Benefit Fraud, and Identity Theft Offenses
25
PBS NewsHour. Former Election Official Fact-Checks Trump’s Claims of Election Fraud in California
26
Cornell Law Institute. 52 U.S. Code § 20511
27
FBI. Election Crimes and Security
28
Justia. Crawford v. Marion County Election Board, 553 U.S. 181
29
MIT Election Data and Science Lab. Voter Identification
30
National Conference of State Legislatures. 9 Things to Know About the Proposed SAVE America Act
31
MIT Election Data and Science Lab. New Report: How We Voted in 2024
32
MIT Election Data and Science Lab. How We Voted in 2024