How many documented cases of wrongful deportation of U.S. citizens have been verified by federal audits since 2000?
Executive summary
Federal audits and watchdog analyses have verified a measurable but limited number of wrongful removals: government and third‑party analyses identify at least 70 U.S. citizens who were deported in error during the period reviewed by auditors, and audits have repeatedly concluded the agencies lack reliable systems to count and prevent such mistakes [1] [2] [3]. Available audited findings therefore confirm dozens of verified wrongful removals since 2000, but they also stress that the true total is unknowable because ICE and CBP do not systematically track citizenship investigations [2] [1].
1. What the federal audits actually counted
The Government Accountability Office (GAO) reviewed immigration enforcement records and processes and found evidence that ICE and CBP had taken enforcement actions against people who later were determined to be U.S. citizens, concluding that agency data and guidance were inconsistent and incomplete for citizenship investigations [2]. Independent analyses that drew on agency data—most prominently Transactional Records Access Clearinghouse (TRAC) analyses cited by the American Immigration Council—found ICE records showing 70 removals of people identified as U.S. citizens during 2002–2017; auditors and watchdogs treat that figure as the audited, documented minimum of wrongful removals in the period reviewed [1] [3].
2. Why “at least 70” is the defensible audited figure
Multiple oversight products and reporting converge on “at least 70” wrongful removals as the number the government data can substantiate: TRAC’s analysis of ICE records produced the 70 removals figure and GAO’s enforcement data review flagged the same subset of removals as a verified minimum while recommending mandatory tracking to close gaps [1] [3] [2]. In other words, federal audits and the best-available data do verify dozens of cases—but they stop short of producing a comprehensive tally because the underlying systems do not record citizenship outcomes reliably [2].
3. The critical caveat: undercounting and data blindness
GAO’s central finding is that ICE and CBP policies sometimes require documenting citizenship investigations but do not require updating citizenship status in their systems once evidence of U.S. citizenship is found, creating systemic blind spots and meaning any number derived from existing records is a floor, not a ceiling [2]. Reporting by watchdogs and journalists—such as ProPublica’s later tallies of more than 170 Americans held by immigration agents and academic clinics documenting individual wrongful deportations—underscore the likelihood of many uncounted cases and the human consequences of the agencies’ data gaps [4] [5].
4. Competing narratives and institutional incentives
Advocacy groups and civil‑rights organizations emphasize that documented cases are only the tip of the iceberg and press for mandatory tracking, remedies and accountability, citing audited evidence and individual lawsuits as proof systemic change is needed [6] [7]. The agencies, while acknowledging policy weaknesses in guidance and systems, have not historically maintained or released a comprehensive, audited accounting of all cases, meaning congressional and GAO pressure has been required to surface the documented minimum and to demand reforms [2] [7]. Readers should note that some later media and advocacy claims cite broader tallies of citizens detained or “wrongfully affected” beyond removals; those counts are important context but are not the same as audited, verified deportations in GAO/ICE data [4] [1].
Bottom line
Federal audits and the best publicly available government data verify at least 70 documented wrongful removals of people identified as U.S. citizens in the period covered by the audits (notably 2002–2017), but GAO and independent investigators make clear this figure is a minimum because ICE and CBP do not systematically track or update citizenship status, so the true number since 2000 cannot be determined from existing audited records [1] [2] [3].