How do states detect and verify noncitizen registrations and what are the error rates in those systems?
Executive summary
States detect possible noncitizen voter registrations by cross-checking voter rolls against federal and state databases, using tools like USCIS’s SAVE service and state DMV/SSA matches, plus manual audits and newer algorithmic forensics; these systems catch very few confirmed noncitizen votes but generate nontrivial false positives that have led to eligible U.S. citizens being flagged or even removed [1] [2] [3] [4]. There is no single, reliable national “error rate” published — available state audits show confirmed noncitizen voting is vanishingly rare (fractions of a percent or explicit counts in the single- or low-double digits), while mismatches and wrongful flags occur at rates large enough to generate litigation and concern [5] [3] [4].
1. How states look for noncitizens: database cross‑matches and SAVE
Election officials typically compare the information on voter registration forms (name, birthdate, SSN or last four digits, driver’s license) with records held by state DMVs, the Social Security Administration, and federal immigration systems; a formal channel many states use is USCIS’s Systematic Alien Verification for Entitlements (SAVE), which returns point‑in‑time citizenship or immigration-status matches based on queries from those source systems [2] [1]. SAVE has been optimized to accept bulk queries and partial SSNs and, as of 2025, eliminated transaction charges and expanded capabilities — leading more states to use it for voter-list maintenance [1] [6].
2. Triage: pending status, outreach, and cancellation rules
When a match fails or returns ambiguous results, states typically move the registration to a “pending” status and notify the registrant to provide documentary proof of citizenship; policy differs by state on whether a SAVE flag alone justifies removal, and federal guidance and lawsuits underscore that flagged voters should receive notice and an opportunity to cure before purging [7] [4] [8]. Several states have pursued faster removals or pushed for documentary proof-of-citizenship laws, prompting pushback that such measures risk disenfranchising lawful voters [2] [8].
3. What confirmed cases look like — very rare, numerically small
Independent reviews and state audits repeatedly find confirmed noncitizen voting incidents are extremely rare: examples include a Michigan probe that found 16 alleged instances among more than 5.7 million ballots (0.00028%), audits in Georgia that blocked roughly 1,600 attempts across 25 years with none succeeding, and a study across 42 jurisdictions that identified about 30 potential incidents in more than 23 million votes — all indicating confirmed noncitizen voting is minute compared with total turnout [3] [9] [5].
4. Error rates and false positives: a persistent and consequential problem
While confirmed noncitizen votes are uncommon, mismatches and false positives are frequent enough to matter: election officials have repeatedly found that many flagged records are actually U.S. citizens or outdated registrations, for example Michigan’s finding that some flagged people were citizens or had canceled registrations, Texas reporting cases like Anthony Nel being wrongly identified by SAVE, and large inactivations in Alabama that later proved to include thousands of eligible voters [3] [4] [10]. Reporting and litigation suggest SAVE and cross‑matches can misclassify people because data sources were collected for different purposes and are sometimes outdated [4] [1].
5. New tools, academic forensics, and the political stakes
Researchers are developing algorithmic and machine‑learning tools to monitor roll integrity and detect anomalies, but those tools depend on daily, clean data feeds that vary by county and state; moreover, political actors pushing nationalized list‑maintenance or expanded access to voter files create incentives that can bias how states use verification data and raise concerns about a “data grab” that could amplify removal errors — a point raised by former DOJ lawyers and civil‑rights groups challenging federal requests [11] [8] [4]. Available reporting does not produce a single, defensible nationwide error percentage; instead, the evidence shows two truths simultaneously: confirmed noncitizen voting is vanishingly rare, and verification systems produce enough false positives to trigger lawsuits, policy fights, and calls for tighter safeguards [5] [4] [3].