Trump's noncitizen voter data remains a mystery. But the claims raise midterm fears

DHS Preliminary Analysis Flags Potential Noncitizens on Voter Rolls in Four States; States Request Methodology Details

Secretary of Homeland Security Markwayne Mullin stated on July 17 that a Department of Homeland Security review identified more than 250,000 non-U.S. citizens on voter rolls across California, Nevada, New Jersey, and Pennsylvania. The four states have disputed the figures and requested additional information on the data sources and methods used.

State election officials in at least three of the states responded to letters from Mullin by the July deadline, questioning the accuracy of the counts and asking for the specific methodology. The Department of Homeland Security has not released a detailed explanation of how the numbers were derived or provided the names of the individuals identified.

Public voter registration lists do not contain Social Security numbers. The DHS letters referenced a subset of 118,003 records in the four states where name, date of birth, address, and Social Security number matched entries in agency files. Nevada Secretary of State Cisco Aguilar, a Democrat, noted that the Department of Justice has an active lawsuit seeking voter roll data including partial Social Security numbers and driver's license information from the state. Similar suits against California, New Jersey, and Pennsylvania were dismissed by federal judges. Aguilar asked how the agency obtained the matching data.

Election officials in the affected states have stated that they cannot verify or act on the claims without the underlying records. California Secretary of State Shirley Weber cited state law restricting the sharing of voters' sensitive information and expressed doubts about the data's reliability. Nevada and Pennsylvania officials similarly requested descriptions of the analysis process, including whether commercial data brokers or artificial intelligence tools were involved.

New Jersey officials separately reported discovering thousands of noncitizens on the state's voter rolls, attributing the entries to a software glitch and stating that an investigation is underway. This disclosure occurred around the same period as the DHS letters.

Multiple prior reviews by state and federal agencies have concluded that voting by noncitizens in federal elections occurs at low rates. A 2024 statement from U.S. Citizenship and Immigration Services described such voting as extremely uncommon while noting that existing laws address the issue. The Heritage Foundation maintains a database documenting 99 cases of noncitizen voting since 2000. Data-matching efforts that lack unique identifiers such as Social Security numbers have produced high rates of false positives in past attempts, according to election data specialists.

The National Voter Registration Act prohibits systematic removal of registrations in the 90 days before a federal election. That period for the November midterms begins August 5. State responses to the DHS letters cited this restriction and emphasized the need for due process to avoid removing eligible voters.

DHS letters described the analysis as a preliminary review and offered resources for identity verification. Mullin stated during the July 17 press conference that additional reviews would occur before, during, and after the election. A federal judge in Washington, D.C., halted use of an expanded version of the DHS SAVE system in late June, finding its creation unlawful and its outputs potentially inaccurate; the administration is appealing that decision.

The Department of Justice has pursued lawsuits against more than two dozen states, including the four named, seeking voter roll data with personal identifiers. Those cases remain pending or have been dismissed depending on the jurisdiction.

David Becker of the Center for Election Innovation & Research stated that without transparent methodology and individual records, the figures cannot be evaluated. State officials have indicated they will not remove voters based on the preliminary numbers alone.

The letters and responses leave unresolved how the agency compiled the 250,000 figure or the smaller matched subset. Election administrators in the four states continue to seek the specific voter data referenced in the communications.