Two weeks after the Department of Homeland Security announced it had identified more than 250,000 noncitizens on voter rolls across four states, the officials responsible for those very rolls say they still have no idea where the number came from. That gap, between a dramatic federal claim and the total absence of underlying data to support it, has left state election officials frustrated and, in some cases, genuinely alarmed about what the claim might be used for as November's midterms approach.

A number with no visible math behind it

The announcement came from Secretary of Homeland Security Markwayne Mullin at a July 17 press conference, where he alleged that voter roll analysis had turned up more than 250,000 noncitizens registered to vote across California, Nevada, New Jersey, and Pennsylvania. Mullin promised the announcement was only the beginning, telling reporters "we will pursue maximum pressure on this" and signaling additional similar findings would follow in the weeks ahead.

What Mullin did not offer, then or in the two weeks since, is any explanation of how DHS arrived at that specific figure. According to reporting from NPR, DHS asked the four affected states to formally respond to its allegations by Thursday, yet the agency itself has not disclosed the methodology it used to analyze state voter lists, nor has it provided the underlying data that would let those states independently verify or dispute the claim.

Officials asking for basic transparency and getting little back

Nevada Secretary of State Francisco Aguilar captured the core frustration succinctly, noting that DHS has not even provided him with the names of the Nevada registered voters the agency claims may be noncitizens. "Until I actually see the data, I can't do anything about it," Aguilar said, a comment that underscores just how little state officials currently have to work with despite being asked to respond to a serious and highly public allegation about the integrity of their own voter rolls.

Secretaries of state from the affected states have collectively pushed back, demanding detailed descriptions of the methodology DHS used, along with transparency about exactly which state voter lists the agency analyzed. California and Nevada officials went further, specifically asking whether DHS had used artificial intelligence tools as part of its analysis, a question that remains unanswered and highlights how much uncertainty still surrounds the process behind the headline number.

Why researchers are skeptical from the outset

The claim also runs against a substantial body of existing research on the underlying question. Several independent reviews conducted in recent years have consistently found that actual voting by noncitizens in U.S. elections is extremely rare, a pattern documented across multiple studies examining different states and election cycles. That existing research doesn't prove DHS's specific 250,000 figure is wrong, since a name appearing on a voter roll is a different question from that person actually casting a ballot, but it does mean the scale of DHS's claim runs well ahead of what prior nonpartisan research on this exact question has typically found.

The fear driving state officials' concern

Beyond the immediate dispute over data and methodology, multiple state officials told NPR they worry about a more consequential possibility: that these unverified numbers, or similar figures unveiled closer to the general election, could eventually be used as a pretext for warrants allowing federal law enforcement to intervene directly in vote-counting processes. With Election Day now fewer than 100 days away at the time these concerns were raised, that timing consideration adds real weight to officials' worry that an unsubstantiated claim introduced now could still shape how the federal government justifies intervention later in the election cycle, even if the specific 250,000 figure is never fully verified or explained.

A dispute with implications well beyond four states

Even though DHS's analysis reportedly covers only California, Nevada, New Jersey, and Pennsylvania, the controversy's shadow extends further. Mullin's promise of additional similar announcements in the coming weeks means other states could soon find themselves in the same position these four currently occupy: publicly accused of harboring a specific, large number of noncitizen voters, without access to the data or methodology needed to confirm, refute, or correct that claim before it shapes public perception of the upcoming midterms.

That dynamic puts affected secretaries of state in a genuinely difficult position. They can neither confirm the allegation, since they lack the underlying data DHS is using, nor definitively refute it, for the same reason. Their only immediate option is what they have already pursued: demanding the transparency needed to evaluate the claim on its actual merits, a request DHS has so far declined to fully satisfy.

What resolution, if any, looks like from here

With DHS's own Thursday deadline for state responses now passed and the agency still declining to detail its methodology, the immediate dispute appears likely to remain unresolved for the foreseeable future. Whether DHS eventually releases the specific data and analytical process behind its 250,000 figure, whether additional states receive similar allegations in the coming weeks as Mullin has signaled, and whether any of these claims translate into concrete federal action around the midterms are all questions that remain genuinely open. For now, the officials most directly responsible for the accuracy of America's voter rolls say they are being asked to answer for numbers they cannot see, verify, or trace back to any disclosed source.

---

*Sources cited in this article include reporting from NPR and its member stations, including coverage published July 31, 2026, featuring statements from Homeland Security Secretary Markwayne Mullin and Nevada Secretary of State Francisco Aguilar. This article presents the Department of Homeland Security's claims alongside the affected states' response; all figures reflect reporting available as of July 31, 2026.*