The Numbers Behind America’s Noncitizen Voting Fight

The fight over noncitizen voting in the United States is less about actual illegal ballots than about how political actors weaponize imperfect data to intimidate election officials and reshape the rules of participation.

Key Points

DHS’s headline claim of 250,000 “noncitizens on voter rolls” rests on unverified name-matching and has not been backed by released methodology or proof of illegal voting.
Independent audits and multi‑state reviews consistently find noncitizen voting to be statistically negligible, with incidence rates measured in ten‑thousandths of a percent.
Keith Ellison and other state officials frame the SAVE Act and related federal threats as an abuse of power that would criminalize routine election work and erect barriers for eligible voters.
The deeper pattern is familiar: large, dramatic numbers based on preliminary data, followed by forensic review that collapses those claims to tiny, non‑impactful figures.

How DHS Got to “250,000 Noncitizens” — And What That Number Really Is

When Homeland Security Secretary Markwayne Mullin announced that DHS had identified “over 250,000” noncitizens registered to vote in California, New Jersey, Nevada, and Pennsylvania, it sounded like a bombshell. The figure came from a review of public voter files in states that had not integrated DHS’s SAVE (Systematic Alien Verification for Entitlements) database, matched against federal immigration records. In the letter to Pennsylvania, for example, DHS said there “may be as many as” 14,576 noncitizens on the rolls, but acknowledged that only 8,594 were direct matches in its own files. Even in the administration’s telling, then, the number is not a confirmed count of illegal voters; it is a speculative upper bound derived from name and data matching in imperfect systems.

Crucially, DHS has not publicly released the underlying dataset, its matching rules, or any audit showing how many of the flagged individuals are in fact noncitizens. Nor has the department said how many of these registrants ever cast a ballot. Independent reporters and fact‑checkers who scrutinized the claim found that it relies on two opaque searches, not on a transparent methodology that outside analysts can replicate or test. The headline number is therefore best understood as a preliminary screening result — the kind that professionals treat as a starting point for verification, not as proof of widespread fraud.

What the Evidence Shows About Actual Noncitizen Voting

To assess whether Mullin’s warning reflects a genuine threat to election integrity, it is necessary to separate registration anomalies from ballots cast and to look at broader empirical work rather than isolated talking points. That wider record is unusually consistent. The Bipartisan Policy Center, a centrist think tank, reviewed state investigations and concluded there is “no evidence that noncitizen voting has ever been significant enough to impact an election’s outcome.” The Center for Election Innovation & Research, which has tracked state disclosures about noncitizens on voter rolls across the country, similarly found that the numbers are “extremely small” and pose no systemic issue.

Concrete state audits bear this out. A Michigan review of its entire voter roll found 16 credible instances of noncitizen voting in a general election, an incidence rate on the order of 0.00028% of ballots cast. A Georgia audit cited in multistate correspondence reported 20 noncitizens out of 8.2 million voters. Earlier work examining the 2016 presidential election across jurisdictions encompassing 23.5 million votes identified only about 30 suspected noncitizen voting cases — roughly 0.0001%. These are not numbers that can plausibly swing outcomes. They are, instead, the kind of statistical noise that every large administrative system generates.

Organizations across the ideological spectrum — from the Brennan Center for Justice to the American Immigration Council — have examined alleged noncitizen voting and reached the same conclusion: incidents do occur, usually through clerical error or misunderstanding, but systematic fraud by noncitizens is “vanishingly rare.” Federal law already prohibits noncitizens from voting in federal elections and imposes serious penalties; voters must attest under penalty of perjury that they are citizens when they register. When states have conducted targeted investigations in response to sensational claims, the pattern has been repetitive: large allegations based on database matches, followed by a handful of confirmed mistakes and almost no intentional illegal voting.

The SAVE Program: Imperfect Tool, High Political Stakes

DHS’s push to use the SAVE system as a standard tool for cleaning voter rolls is central to this controversy. SAVE was designed to help agencies verify eligibility for public benefits by checking immigration and citizenship records. Applied to voter registration, it compares state lists against federal data to flag possible noncitizens. In theory, that sounds straightforward. In practice, it is fraught. Citizenship status changes over time; names are shared and misspelled; federal records can lag reality. States that have run their voter rolls through SAVE have repeatedly found that many “noncitizen” flags were simply wrong.

Early multi‑state runs of SAVE against voter registrations identified roughly 4,200 apparent noncitizens among about 35 million registered voters — about 0.01%. Subsequent investigation showed that a substantial share of those were citizens misclassified because the federal database had not been updated or could not match records correctly. Minnesota’s Secretary of State, in a formal comment to DHS, reported that state audits found a suspected noncitizen incidence rate of just 0.00028%, underscoring how tiny the problem is when records are properly verified.

Yet Mullin and the Trump administration have treated SAVE outputs as a basis for threats. Mullin has warned state election officials that if they do not act on DHS information to “scrub” their rolls, they could face fines, penalties, and up to five years in prison. He has also signaled that access to federal election grants will be conditioned on adopting federal security directives, including SAVE‑based verification. This is not a neutral technical recommendation; it is an attempt to leverage an imperfect tool into federal control over how states manage the most basic aspect of their elections: who is allowed on their lists.

Keith Ellison’s Opposition: Barriers to Voting and Criminalizing Mistakes

Minnesota Attorney General Keith Ellison has emerged as a leading legal voice against this approach. In public analyses of the SAVE Act and related executive initiatives, Ellison argues that the federal plan would “create substantial barriers to voting” by effectively demanding passport‑level documentation from ordinary citizens and by treating routine administrative mistakes as potential crimes. The bill and its implementing directives, in his view, would turn local election officials into targets, threatening them with up to five years’ imprisonment for errors in list maintenance.

Ellison’s concerns are grounded in both law and empirical evidence. Election administration is constitutionally a state function; Congress sets dates and broad rules, but the day‑to‑day mechanics — registration systems, list maintenance, ballot processing — are run by states and localities. Federal civil‑rights enforcement exists to prevent discrimination, not to micromanage list‑cleaning. Courts have repeatedly rejected attempts by the Justice Department to seize sensitive voter data or impose federal proof‑of‑citizenship rules, handing DOJ a 16–0 loss record in such cases. Ellison and allied attorneys general successfully obtained a preliminary injunction blocking an executive order that would have required documentary proof of citizenship to register and curtailed mail‑in voting.

From Ellison’s vantage point, Mullin’s threats fit a broader strategy: exaggerate an empirically tiny problem, use that narrative to justify stringent documentation requirements and federal audits, and in the process discourage participation by eligible voters who lack passports or who fear making mistakes. When the incidence of confirmed noncitizen voting is measured in hundredths of a basis point, sweeping new barriers aimed at this issue look less like targeted security measures and more like voter suppression by another name.

State Officials Push Back: Intimidation vs. Partnership

Ellison is not alone in treating DHS’s posture as abusive rather than protective. Secretaries of state from Arizona to Colorado have publicly rebuked Mullin’s threats, emphasizing the robustness of existing safeguards and the corrosive impact of federal intimidation on the workforce that actually runs elections. Adrian Fontes, Arizona’s secretary of state, described Mullin as ignorant about election administration and pointed to the extensive checks already in place: certification of officials, logic and accuracy tests on every tabulator, and rigorous post‑election audits.[MS NOW: Fontes transcript] In Colorado, Secretary of State Janet Griswold stated bluntly that she would “not bend to intimidation” and testified to receiving thousands of threats since 2021, a climate she traces directly to politicized attacks on election integrity.[MS NOW: Griswold transcript]

Election security professionals share that concern. Former CISA director Chris Krebs, who oversaw federal election security in 2020, has emphasized that the infrastructure built through cooperative briefings, shared threat intelligence, and technical support has been undermined by the shift to top‑down directives and public threats of prosecution.[Face the Nation: Becker/Krebs transcript] David Becker, a longtime election lawyer, notes that “the total number of election officials who’ve been arrested for wrongdoing is zero” and that states generally have better, more current data on citizenship than the federal SAVE system.[Face the Nation transcript] In their view, Mullin’s approach replaces partnership with coercion, and evidence‑based risk management with headlines aimed at political gain.

The Larger Pattern: Imperfect Rolls, Inflated Claims, Negligible Impact

To understand why this clash keeps recurring, you have to look beyond today’s numbers. Voter registration lists in a country as large and mobile as the United States are inherently imperfect. People move, change names, gain citizenship, or die; databases are updated in batches, not continuously. When you run massive cross‑checks between such systems, you inevitably generate thousands of “matches” that are not what they seem — an outdated record, a shared name, a naturalized citizen whose federal file was never updated. Analysts who specialize in election data, like the Center for Election Innovation & Research, have found that sweeping allegations about noncitizen registrations almost always arise from misunderstandings or mischaracterizations of this complexity.

The pattern has played out repeatedly. In 2016, claims that millions of “illegals” were poised to vote were traced back to a decade‑old survey with widely criticized methodology, then inflated by political campaigns to their highest speculative estimates. More recent assertions of hundreds of thousands or even millions of noncitizen voters have collapsed under scrutiny to small handfuls of cases, many of them inadvertent registrations at motor vehicle offices. Each time, the empirical bottom line is the same: noncitizens do appear on voter rolls in small numbers, and a few may vote, but the rates are far too low to matter for outcomes and far too scattered to support the story of coordinated, large‑scale fraud.

In that context, DHS’s 250,000 figure looks less like a breakthrough discovery and more like the latest iteration of a familiar narrative: a large, unverified number raised in a charged political moment, paired with threats of prosecution and promises of new restrictive legislation, and lacking the transparent evidence that would allow experts to audit the claim. When independent auditors finally get access to specific records, the history suggests they will find far fewer true noncitizens — and even fewer illegal ballots — than the headline implies.

Step four: use threats and intimidation to try to get your way. Friday, Secretary of Homeland Security Mullin threatened state election officials with possible prison time if they don’t comply with Trump administration efforts to determine whether noncitizens are on voting lists.

— Jane Doe’s Addiction (@Jane_Plain_Doe) July 19, 2026

What This Means Going Forward

The stakes in this debate are not abstract. When federal officials threaten to “get ’em all” — not just the rare noncitizens who knowingly break the law, but also the election workers and state officials who manage imperfect lists — they change who is willing to serve and how aggressively states purge their rolls. That, in turn, affects eligible voters: naturalized citizens whose status is misrecorded, long‑time residents who lack passports, and ordinary people who make honest mistakes in paperwork.

For readers trying to make sense of competing claims, the most reliable guide is the weight of evidence, not the volume of rhetoric. On that score, the record is clear. Noncitizen voting exists, but at rates so low that they do not threaten election outcomes. The SAVE program and other databases can help states improve list accuracy, but they are not precise instruments; used crudely and under threat of criminal sanctions, they risk doing more damage than good. And officials like Keith Ellison are right to warn that broad, punitive federal interventions grounded in speculative numbers are less about protecting democracy than about redefining who gets to participate in it.