The American Immigration Council is a non-profit, non-partisan organization. For more than a decade, some politicians have called for substantial changes to how Americans vote in federal elections, including strict voter ID and additional citizenship screening during voter registration, based on claims of noncitizen voting. President Donald Trump has demanded that Congress pass the Safeguard American Voter Eligibility Act, also known as the SAVE America Act, which would mandate specific photo ID procedures during voting, require that Americans provide proof of citizenship when registering to vote, and mandate ongoing citizenship verification.
President Trump has repeatedly claimed that huge numbers of “illegal aliens” are voting in the United States and has said that failure to pass new elections laws is an existential threat to the country. Despite these claims, available evidence on the issue demonstrates that noncitizen voting is extraordinarily rare — including evidence put forward by one of the country’s foremost proponents of changes to election laws.
The Heritage Foundation has long supported these changes to American voting laws. The conservative think tank also maintains a national database of proven cases of what they describe as “voter fraud,” which they have touted as containing “plenty of proof” that “reforms like the SAVE America Act” are needed.
However, a close analysis of the Heritage Foundation’s own database demonstrates that there is no crisis of noncitizen voting. Even in Heritage’s own data, cases of noncitizen voting are exceedingly rare; further analysis of these cases shows that most involve “voter mistake” and not “voter fraud,” such as situations where a person unintentionally violated an election law.
The Heritage Foundation’s data makes clear that while noncitizen voting does happen in extraordinarily rare occasions, it is not a widespread problem impacting the outcome of any elections. Cases of outright voter fraud by noncitizens are exceptionally rare to the point of being a statistical nullity.
What do these cases tell us? First, in 2024, the American Immigration Council analyzed this database and found that it contained just 68 cases of proven noncitizen voting dating back to 2000, amounting to less than 0.0001% of total votes. Since 2024, the Heritage Foundation’s database has grown, with an additional 21 cases added. In total, noncitizen voting makes up just 5.5% of Heritage’s broader database of 1,619 cases of election-related offenses.
These 89 people include cases from around the United States. In the seven presidential elections carried out over that period, the number of people voting ranged from a low of 105 million in 2000 to a high of 158 million in 2020. Even if every one of these noncitizens had voted in the same election, it would still represent less than 0.0001% of total votes.
Other analyses of noncitizen voting produce similar results. A 2026 analysis by Reuters of noncitizen voting cases brought in federal court under a 1996 law found that over the last 30 years, just 129 people have ever been criminally charged in federal court for “voting by aliens,” and only 73 of those cases resulted in a conviction.
Second, the Heritage Foundation data also shows that the most common scenario involving noncitizen voting is not an “illegal alien” conspiring to fraudulently vote; it is a lawful permanent resident (a green card holder) registering and voting out of a mistaken belief that they were permitted to do so.
For example, in 2002, the Bush administration prosecuted 15 green card holders in Florida for unlawful voting. In 11 out of 15 of those cases, the defendants admitted to voting but said they were unaware that people with green cards couldn’t vote or had been told by someone else that they were allowed to vote.
This conclusion is also supported by the Reuters analysis, which showed that out of the 73 federal convictions for noncitizen voting since 1996, nearly 80% involved lawful permanent residents. An August 2026 audit in Michigan which led to the prosecution of 5 individuals for noncitizen voting showed the same result; 3 out of 5 people charged had green cards, and the other two had lawful but temporary status.
Third, the Heritage Foundation data shows that in the even more rare cases of undocumented immigrants voting (just 16 cases in total), almost all involved individuals had adopted false identities as U.S. citizens using fake or stolen documentation.
In each case, at the time the person voted, the government had no reason not to believe the person was a U.S. citizen — because they had the documents to prove it. These rare cases would not have been prevented even if the person had been asked to present voter ID or go through a pre-registration verification of citizenship.
Fourth, the data shows that government error is often responsible for noncitizen registration and voting. The Heritage Foundation’s data is not comprehensive. Further cases not included in Heritage’s database reveal at least 19 additional cases of convictions for noncitizen voting that are not included in the database. Of those, at least 16 involved lawful permanent residents, including a Panamanian grandmother who came to the United States legally in 1973 and who was registered to vote in 2016 despite presenting a driver’s license to state officials which indicated she was not a citizen.
This type of error is typical in cases of noncitizen voting. Multiple people included in Heritage’s database were registered to vote despite telling government officials they were not citizens. Recently, the State of New Jersey admitted that a software error had led to over 6,600 people being inadvertently registered to vote despite having checked a box affirming that they were not U.S. citizens.
New Jersey officials say that fewer than 400 of those people eventually voted using that registration. While it’s unknown how many of those people had become naturalized citizens by the time they voted, even if all 400 were still noncitizens, that would represent just 0.009% of total votes in the 2024 elections; under one hundredth of one percent.
Government error is also often responsible for erroneous claims of high numbers of noncitizen voting. The Department of Homeland Security’s SAVE program, which the Trump administration has sought to require states to use, is known to produce false positives. When state officials recently used the database in Texas to identify suspected cases of noncitizen voting, it flagged multiple U.S. citizens by mistake. In one use of the SAVE program in Missouri, fully 35% of potential noncitizens flagged by the data were naturalized citizens at the time they voted.
Preventing the rare and exceptional instances of noncitizen voting would be better achieved through policies that offer better training for government workers and more careful procedures to avoid errors like those that occurred at the hands of state administrators in New Jersey.
Given the statistical rarity of noncitizen voting, efforts to deploy ICE agents or federal officers to polling places would likely do more to suppress American voters than stop any noncitizen from voting. Any debates around changes to election laws should be grounded in the facts, not in hyperbole.
