FACT-CHECK: The Myth of Non-Citizens Voting in Droves—Debunked with Supporting Data

Percentages that are virtually zero

Let’s start with the most comprehensive data available. In 2016, the Brennan Center conducted a national survey of 44 election officials representing 42 jurisdictions in 12 states—including 8 of the 10 jurisdictions with the highest number of non-citizens nationwide. These officials had overseen the counting of 23.5 million votes. The result: they reported only 30 suspected incidents of voting by non-citizens for investigation or prosecution. That is 0.0001% of the votes cast. In other words: one incident for every 783,000 ballots. Forty of the 42 jurisdictions reported no known incidents.

After the 2016 election, The New York Times interviewed officials from 49 states and the District of Columbia. It learned of two possible cases of non-citizen voting—out of 137.7 million voters. The Washington Post, for its part, identified four proven cases of election fraud across all categories, and no cases of non-citizens voting. News21, an independent journalism organization, documented 56 alleged cases of non-citizen voting over the entire 2000–2012 period. If we were to extrapolate, assuming that all these cases were true and that all these people had voted in 2016, this would represent approximately 0.00004% of all ballots cast.

The Heritage Foundation’s Legacy Backfires

The Heritage Foundation, a conservative organization that maintains a database dedicated to documenting U.S. election fraud—a database designed specifically to make fraud appear to be a systemic problem—has recorded only 68 confirmed cases of non-citizens voting over the past four decades, dating back to the 1980s. That represents less than 5% of all entries in their database, which covers all forms of election fraud combined. Out of more than one billion votes cast during this period in thousands of elections, the incidence of proven cases of non-citizen voting is less than 0.0001%. Of these 68 cases, only 10 involved undocumented immigrants. The American Immigration Council, which conducted this detailed analysis, points out that even an organization whose mission is to document fraud has been able to demonstrate only an infinitesimal amount of it.

There is something intellectually honest about turning to the Heritage Foundation to debunk the argument of widespread fraud. When even your most determined allies, using their own database designed to show the scale of the problem, can’t find more than 68 cases in 40 years out of a billion votes—the debate is over. The numbers don’t lie.

Wave after wave of investigations that turned up nothing

Over the years, whenever a state has conducted a serious investigation into voter fraud by non-citizens, the results have been the same: virtually nothing. In 2013 and 2015, Ohio identified 44 non-citizens who had voted in at least one election since 2000. Secretary of State John Husted himself stated: “None of these votes affected the outcome of an election.” In 2017, Ohio reported 126 non-citizens who had voted since 2013—representing about 0.0015% of the roughly 8 million people who vote in the state in each election. In 2020, a new investigation uncovered 13 cases of alleged voting by non-citizens—representing 0.00016% of the ballots cast in that year’s primaries and general elections.

Colorado in 2012 provides a textbook example of the typical pattern: the secretary of state initially claimed that up to 11,805 non-citizens were registered to vote. After a thorough investigation, 35 individuals were identified as having potentially voted. In Nevada in 2016, out of more than 1.1 million ballots cast, the audit identified 3 non-citizens who had voted—or 0.0003%. In North Carolina over the 2002–2012 period, with 19.5 million votes cast, 58 potential cases of voting by non-citizens were referred to prosecutors—representing 0.0003% if all allegations are accepted as true.

The Mechanics of False Positives

A recurring phenomenon in these investigations is worth noting: the tendency for initial announcements to vastly exceed the actual results after verification. In Texas in 2019, approximately 100,000 voters were initially flagged as “possible non-citizens” by the Secretary of State—an announcement that made headlines. A federal court blocked the voter purges: most of these individuals were in fact naturalized citizens whose data had not been properly cross-checked. In 2012, Florida reviewed 180,000 potentially non-citizen voters, removed 85 from the rolls, and ended up with a single conviction. Michigan in 2012: the Secretary of State had claimed that up to 4,000 non-citizens were registered. The result of the investigation: 10 individuals were referred to the attorney general.

This pattern of inflated claims followed by a collapse of the actual figures upon verification is characteristic of propaganda, not of evidence-based public policy. Each wave of investigations led by Republicans has confirmed what Democrats and independent researchers have been saying for years: the problem does not exist on the scale claimed. And yet, the rhetoric remains unchanged.

Existential risks for no benefit

Beyond the statistics, there is a fundamental logic that explains the rarity of this phenomenon. Voting illegally as a non-citizen in the United States exposes one to extremely severe penalties. Under federal law in effect since 1996, a non-citizen who votes in a federal election faces up to five years in federal prison. In addition to the prison sentence, the individual faces deportation, which means expulsion from U.S. territory. Furthermore, for a legal resident in the process of naturalization, voting illegally can permanently nullify any prospect of becoming a U.S. citizen. The Migration Policy Institute sums it up unequivocally: non-citizens face the risks of federal imprisonment, deportation, and revocation of their legal status—even if no vote is cast, simply for having checked the registration box.

Faced with these existential risks, what would be the benefit? A single vote, in a nation of 330 million people, in an election where the margins of victory in major national contests are measured in millions of votes. The most basic rationality strongly discourages such an act. Moreover, as the Brennan Center points out, many undocumented immigrants avoid any contact with government institutions for reasons of personal safety. Registering on official voter rolls—with their name, address, and personal information—would represent a considerable and deliberate risk for a benefit close to zero.

The existing verification system works

U.S. law already includes multiple safeguards. Federal registration forms require applicants to declare under oath that they are U.S. citizens, under penalty of prosecution for perjury. The original SAVE program—before it was modified by the Trump administration—already allowed states to query immigration databases to verify the status of their registrants. Senator Alex Padilla (D-California), a member of the Senate Rules Committee, stated during the June 4, 2026, vote: “The current safeguards work. And yes, it is already illegal for non-citizens to vote in the United States.” U.S. Citizenship and Immigration Services itself acknowledged in an October 2024 letter that the evidence clearly shows these laws “work as intended” and that it is “extremely rare” for non-citizens to vote in federal elections.

The central paradox of the SAVE Act is right here, in black and white: to solve a problem that official government data itself describes as “extremely rare,” the proposal is to eliminate electoral safeguards that allow tens of millions of legitimate citizens to vote. The cure is a hundred times more dangerous than the disease. In a democracy, this is a catastrophic reversal of priorities.

From a Targeted Tool to a Mass Database

The SAVE system—Systematic Alien Verification for Entitlements—was originally designed to allow social welfare agencies to verify the immigration status of benefit applicants. Under the Trump administration, through an executive order signed on March 25, 2025, the system was radically reconfigured. What was once a database containing the immigration information of some 26.5 million people was transformed into a citizenship verification system providing access to the personal information of nearly all Americans. The changes enabled mass searches using partial Social Security numbers and made the database freely accessible to state and local election officials.

The concrete result: the expanded SAVE program was used to verify the status of more than 67 million registered voters, mostly in Republican-led states. The system flagged thousands of voters as “potential non-citizens.” Subsequent investigations showed that many of them were, in fact, U.S. citizens eligible to vote. The combined data included information from the Internal Revenue Service and the Department of Labor, integrated into a “data lake” that allowed for cross-referencing, including biometric data and information on citizenship status. All of this, according to Judge Sooknanan, was based on data that the agencies knew to be unreliable.

The June 22, 2026, Ruling

On June 22, 2026, District Judge Sparkle L. Sooknanan issued her 75-page ruling, the ninth in a series of judicial rejections of the Trump administration’s 2026 election initiatives. The court concluded that the changes made to the SAVE system violated the Social Security Act—which expressly prohibits the disclosure of Social Security numbers—several provisions of the Privacy Act of 1974, and the Administrative Procedure Act. The judge found that federal agencies had “deliberately combined and misused the private information of millions of Americans, including citizenship data they knew to be unreliable,” in their rush to comply with an executive order. She ordered DHS to restore the system to its pre-amendment state. The administration has the option to appeal to the U.S. Court of Appeals for the District of Columbia Circuit.

When a federal judge has to write that the government “haphazardly combined”—clumsily combined, in a disorderly rush—the private data of millions of citizens based on information it itself knew was unreliable, one can gauge the extent of the betrayal. This is not an election strategy. It is an experiment in mass surveillance conducted on Americans, in the name of combating fraud that the numbers have never confirmed.

A Majority That Can’t Be Found, Even Within Trump’s Own Party

On June 4, 2026, the U.S. Senate rejected the SAVE America Act during a marathon “vote-a-rama” on an immigration funding bill. The amendment, sponsored by Republican Senator Lindsey Graham, failed by a vote of 48 to 50—well short of the 60 votes needed to overcome a filibuster. Even more telling: the result did not even reach a simple majority of 50 votes, as four Republican senators joined all the Democrats in voting against it. The Republican dissenters were Susan Collins of Maine, Lisa Murkowski of Alaska, Thom Tillis of North Carolina, and Mitch McConnell of Kentucky. A second amendment, introduced by Senator Mike Lee and limited solely to electoral provisions, failed by a narrower margin, 50 to 49.

This was the bill’s second defeat of the year in the Senate. Prior to that, in March 2026, an attempt at a talking filibuster had also failed. The House, however, had passed the bill on February 11, 2026, by a vote of 218 to 213, with almost exclusive support from Republicans. Donald Trump had described the bill as “common-sense legislation that will save the country” and had declared it a priority over all other legislation. He had threatened never to support any senator who voted against it. Despite this pressure, the anti-SAVE coalition held firm, bolstered by Senator Padilla’s repeated argument: the bill creates a massive problem to solve an infinitesimal one.

The Implications for the 2026 Midterms

The stakes are high. The November 2026 midterm elections are approaching, and the Trump administration’s entire electoral strategy relied in part on implementing new restrictions before that election. With the failure of the SAVE Act in the Senate and the judicial overturning of the expanded SAVE system, two pillars of that strategy have collapsed. Senate Counsel Elizabeth Macdonough had also previously ruled that the SAVE America Act did not meet the criteria for inclusion in a budget reconciliation bill passed by a simple majority. Five states—Arizona, New Hampshire, South Dakota, Utah, and Wyoming—have already adopted proof-of-citizenship requirements for their elections, according to the Brennan Center, though these do not have the nationwide scope of the federal SAVE Act.

I find a glimmer of hope in these four Republican senators—Collins, Murkowski, McConnell, and Tillis—who stood up to Trump’s threats and party pressure. They are not progressive figures. They are pragmatists who looked at the data and said no. In politics, actions speak louder than words. Those four votes protected millions of Americans from a voter purge based on a statistical lie.

Allegations that are constantly refuted, constantly rephrased

The rhetoric of widespread election fraud did not begin with the SAVE Act. It is part of a well-documented cycle: an allegation is made, it is repeated and amplified, investigations are launched, they find nothing significant, and the allegation is rephrased in a different form a few months later. Perhaps the most striking example is that of Rudy Giuliani, who claimed after the 2020 election that at least 40,000 to 50,000 non-citizens had voted in Arizona—“the reality being probably around 250,000.” Giuliani has since been disbarred from the New York Bar for “knowingly making false or dishonest statements.” The Public Interest Legal Foundation published reports in 2016 and 2017 claiming that thousands of non-citizens were registered to vote in Virginia—including the addresses and phone numbers of innocent citizens. Four citizens sued the organization, and its director was required to issue a formal written apology following an out-of-court settlement.

Studies claiming to demonstrate widespread fraud have been systematically refuted. The 2014 Richman study, which claimed that 6.4% of non-citizens had voted in 2008, was completely discredited in 2015: it was based on fundamental methodological errors in the interpretation of survey data. More recently, Just Facts published a study claiming that 10 to 27% of non-citizens—or between 2 and 5 million people—were illegally registered. No government investigation has confirmed these figures. As documented by the Center for Election Innovation and Research in 2026, “allegations sweeping large numbers of non-citizens off the voter rolls consistently turn out to be misleading overestimates of the actual numbers.”

A Well-Known Red Herring

Political scientist Lorraine C. Minnite of Rutgers University has studied allegations of election fraud for over a decade. In an analysis of the first three years of a U.S. Department of Justice initiative launched to detect election fraud, she identified 14 convictions of non-citizens for voting. Fourteen. Over years of targeted federal investigation. Minnite concludes that election fraud, including voting by non-citizens, is “extremely rare.” The finding is decisive: allegations that prove to be baseless far outnumber those that prove to be valid. The machinery of alleged election fraud produces headlines, laws, and purges—but no evidence.

We must call this logic what it is: a diversionary tactic. Creating a fuss over nonexistent fraud serves to obscure the real question—who stands to gain from certain categories of citizens not voting? The answer—which the demographic data from the voter purges confirms time and again—is not hard to find. It is not democracy that is being protected. It is a fragile majority that is being artificially preserved.

Women, rural residents, young people, and minorities

While the SAVE Act officially targeted non-citizen fraudsters, its documented effects would have primarily impacted very specific categories of legitimate U.S. citizens. The Brookings Institution notes that the documentation requirements would have had a disproportionate impact on women in heterosexual marriages, as their documents (birth certificates, passports) may not reflect their current legal name after marriage. The Center for American Progress estimates that approximately 69 million American women who changed their names after marriage may have documents that do not match their current legal names. These women would have been forced to go through additional administrative steps simply to vote.

The Brennan Center identifies other groups that are disproportionately affected: young voters, who often have not yet obtained a passport and whose birth certificates may be difficult to locate; communities of color, particularly naturalized citizens whose naturalization documents may differ from other forms of identification; survivors of natural disasters who have lost their documents; Native Americans, whose civil registration documents follow different systems; and LGBTQ+ individuals, whose names on birth certificates may not match their current identities. According to the Brennan Center, the makeup of the 21 million citizens without easy access to the required documents is representative of the overall U.S. population—but with significant overrepresentation in these specific categories.

A Political and Social Cost for a Non-Existent Benefit

The final comparison is damning. The maximum number of non-citizens who have voted illegally in a U.S. federal election, according to all available studies, is in the tens or a few hundred—never in the thousands, never in the tens of thousands on a national scale. The SAVE America Act, had it been enacted, would have potentially prevented tens of millions of legitimate citizens from voting, according to various estimates. Nonprofit VOTE estimates that a few dozen cases of fraud might have been prevented. According to an analysis cited on YouTube in 2026, the Heritage Foundation’s database—designed to maximize the visibility of fraud—documents only 66 cases over 24 years where the SAVE Act might have made a difference. That’s 2.3 cases per year nationwide.

Two point three cases per year. That’s the real problem. And to solve it, the proposal is to purge 21 million legitimate citizens from the voter rolls. If this equation had been presented honestly, without the smoke and mirrors of grand statements about “electoral integrity,” no one would have gone along with it. That’s why it’s never presented honestly.

Nine Rejections in 2026

According to available data, Judge Sooknanan’s June 22, 2026, ruling marked the ninth time in 2026 that federal courts had rejected the Trump administration’s electoral initiatives. This string of legal defeats is telling in itself. At the same time, the Department of Justice had filed lawsuits against 30 states to compel them to submit their voter registration files so they could be tested in the expanded SAVE system. These efforts were blocked or delayed by court rulings. A separate ruling handed down on the same day, June 22, in Maryland also denied the DOJ access to non-anonymized voter registration files.

The judicial trajectory is significant. The Trump administration’s arguments regarding voter fraud by non-citizens were not merely rejected by Democratic judges—they were rejected on substantive grounds: violations of privacy laws, exceeding statutory authority, and arbitrary and capricious procedures. Judge Sooknanan expressly noted that federal agencies knew the data being used was unreliable, yet proceeded anyway. This is not a matter of method—it is a matter of intent.

New Hampshire: A Preview of the Real-World Consequences

New Hampshire, one of five states that have already adopted proof-of-citizenship requirements, provides empirical evidence of the concrete effects of these policies. According to the Brennan Center, New Hampshire’s new laws have already “led to widespread confusion and the inability of eligible citizens to vote.” A federal ruling on May 29, 2026, compelled New Hampshire to reinstate the option for voters to swear an oath of citizenship when they do not have the required documents—since the law had unlawfully eliminated that option. This specific case illustrates the abstract reality behind the statistics: legitimate citizens—longtime voters—have been prevented from voting by documentation requirements designed to combat a phenomenon that the data does not support.

New Hampshire is the testing ground that reveals the truth. When these laws are enacted, it’s not fraudsters who are stopped—we don’t find any. It’s grandmothers, veterans, students, and newlywed women who show up at a registration office without the right papers and leave without having voted. That is the reality of the SAVE Act. Not non-citizens thwarted. Citizens abandoned.

When Democracy Invokes Democracy to Protect Itself from Itself

The rhetorical framing of the SAVE Act as a measure of “electoral security” and “electoral integrity” warrants critical analysis. In a democratic system, electoral integrity rests on two equally important pillars: the accuracy of the vote (ensuring that only those eligible to vote actually do so) and the universality of the vote (ensuring that all those eligible to vote are able to exercise that right). These two objectives can come into conflict. The question, therefore, is: How proportionate is the response to the identified risk?

The data is unequivocal on the issue of proportionality. The identified risk—a few dozen non-citizens voting per year—is infinitesimal. The proposed response—depriving millions of legitimate citizens of their right to vote—is colossal. In any serious cost-benefit analysis, this imbalance cannot be justified. This is what Professor Vladeck of Georgetown explicitly stated. This is what the Brennan Center has been documenting for years. And this is what four Republican senators finally acknowledged on June 4, 2026, by voting against their own party. The argument of electoral integrity, when used to restrict voting rather than protect it, ceases to be a democratic argument. It becomes its opposite.

Trump, the SAVE Act, and the Strategy of Polarization

Donald Trump presented the SAVE Act as a top priority, declaring during his State of the Union address that it must be passed before any other legislation. He labeled opponents of the bill as people “who want to cheat” and threatened to withdraw his support from any senator who voted against it. This extreme approach to political pressure backfired: it hardened the opposition of moderate Republican senators who refuse to be seen as exploiting the right to vote for partisan purposes. Majority Leader John Thune was forced to acknowledge that the numbers simply weren’t there. An honest assessment of the data by judicial institutions—including those appointed by Republicans in the past—has confirmed at every stage that the problem cited did not justify the proposed remedies.

There is a profound irony in the fact that Trump, who presents himself as America’s defender against its external enemies, has devoted so much energy to a battle against an electoral phantom while real problems—the state of infrastructure, the housing crisis, competitiveness with China—await solutions. The imaginary enemy is more politically useful than the real one. That is the great calculation of demagoguery.

Institutional Trust Eroded from Within

This fact-check reveals something deeper than the mere falsity of a political claim. It reveals a systematic strategy to undermine trust in U.S. electoral institutions—an undermining carried out in the very name of protecting those institutions. Every time a Republican official repeats the claim that non-citizens are voting en masse—without any data to back it up—they erode citizens’ trust in a system that, in fact, works. They create uncertainty where the reality is clear. And this uncertainty, deliberately perpetuated, becomes the breeding ground for restrictive laws justified by fear, not by facts.

Countries that have experienced democratic erosion—Viktor Orbán’s Hungary, Erdoğan’s Turkey, Putin’s Russia in the years leading up to its full-blown authoritarian drift—all began with small deviations from the norm, justified by fabricated emergencies. The use of phantom fraud to justify massive electoral restrictions is not unique to the United States: it is a standard tool of authoritarian populism. The West, to remain the West, must call out these mechanisms when they appear, even in the most robust democracies.

The Importance of Facts in Democratic Debate

In a healthy democratic debate, factual claims must be subject to verification. The free press, independent judicial institutions, and nonpartisan research organizations—the Brennan Center, Brookings, the Center for Election Innovation and Research, and the Migration Policy Institute—have all carried out this work with rigor. Their conclusions are consistent. This fact-check is not the work of a partisan opposition: it reflects an academic, legal, and empirical consensus. When Senator Padilla says that the safeguards are working, he is not engaging in politics. He is stating a fact. When Judge Sooknanan says that the government “knowingly” used data it knew to be unreliable, she is not engaging in politics. She is doing her job.

The truth has unexpected allies. Four Republican senators. A federal judge appointed by Biden who cites the law, not ideology. A consensus among researchers ranging from Rutgers to Georgetown, including the Heritage Foundation’s own databases. When the facts align like this, they are as solid as granite. American democracy is suffering, but it is not without resources. That is what keeps me from falling into utter pessimism.

The Democratic Model Under Global Pressure

What is unfolding in the United States around the SAVE Act and the phantom fraud committed by non-citizens is not an isolated phenomenon. Throughout the Western world, populist parties are using the fear of “foreigners stealing”—jobs, identity, votes—as electoral fuel. The rhetoric is universal; the mechanics are identical. What makes liberal democracies strong is precisely their ability to counter these narratives with facts. U.S. courts have been playing this role with remarkable consistency in 2026—nine rejections of Trump’s electoral initiatives in just a few months. Civil society, voting rights organizations, and independent media are also playing this role.

The West can remain the West only if it defends its fundamental values—universal suffrage, the rule of law, and the separation of powers—with equal vigor against their enemies both within and without. China, Russia, and their allies thrive on the American contradiction between proclaimed ideals and the reality of the policies pursued. Every time the United States attempts to restrict voting by citing nonexistent fraud, it offers its adversaries a caricature of democracy. This is not merely an American problem. It is a vulnerability of the entire Western model.

What Europe Must Take Away

For European democracies watching developments in the United States with growing concern, the lesson from this fact-check is twofold. First, epistemic vigilance: when a claim is repeated often enough and with sufficient confidence, it acquires a social force independent of its truthfulness. The lie about non-citizen voter fraud has been repeated thousands of times, cited in presidential speeches, and used to shape federal law. Systematically refuting it requires a commensurate effort. Second, the strength of checks and balances: the U.S. courts, which Trump has attempted to reform in numerous ways, have resisted with remarkable consistency. The institutions of liberal democracy are not mere ornaments—they are bulwarks. Weakening them is a slippery slope from which there is no easy return.

I conclude this fact-check with a conviction that has remained unchanged for years: democracies do not always die in a single blow. They erode slowly through the accumulation of small deviations from the norm—each individually defensible, but collectively fatal. The SAVE Act and the phantom fraud committed by non-citizens are one such deviation. Identifying them, documenting them, and challenging them with facts—that is the work of those who believe that democracy deserves to be defended with the same fervor as those who seek to erode it.

An Unequivocal Conclusion

Following this fact-check, the conclusion is unequivocal. The claim that non-citizens vote en masse in U.S. elections is statistically false. The available data—drawn from government investigations, independent academic studies, journalistic inquiries, and judicial analyses—all point to the same conclusion: voting by non-citizens accounts for an infinitesimal fraction of the ballots cast, in the range of a few hundredths of a thousandth of a percent. The 23.5 million votes cast in 2016 resulted in 30 reported incidents. Out of 137.7 million voters surveyed by The New York Times, two possible cases were found. The Heritage Foundation, using its own anti-fraud database, has documented only 68 cases in 40 years. The U.S. Department of Justice has prosecuted only 35 non-citizens for illegal voting in 21 years.

Democracy as an Ideal to Be Actively Defended

Meanwhile, the SAVE America Act would have potentially deprived 21 million legitimate U.S. citizens of their right to vote by requiring documents that many do not readily possess. The expanded SAVE system, declared unlawful on June 22, 2026, by Judge Sooknanan, has already led to the wrongful cancellation of voter registrations for legitimate citizens. The proposed solution was objectively and massively disproportionate to the actual problem. Democracy is not a natural good that sustains itself. It requires active vigilance, a commitment to facts, and firm resistance to narratives that divert it from its foundations. This fact-check is a modest contribution to that resistance.

Signed, Maxime Marquette, columnist