How have media investigations and think tanks assessed the composition (criminal vs noncriminal) of ICE arrests and removals since 2024?
Executive summary
Since 2024, media investigations and policy think tanks have converged on a clear pattern: a growing proportion of ICE interior arrests and removals involve people with no recent criminal convictions, even as the agency emphasizes targeting those with criminal histories; the gap between critics and ICE stems largely from different datasets, definitions (administrative vs. criminal arrests), and timeframes used in analyses [1] [2] [3].
1. How ICE describes its own caseload and priorities
ICE’s Enforcement and Removal Operations (ERO) presents dashboards and an annual report that stress the agency’s criminal-enforcement role, reporting that 88,763 of 271,484 ERO removals in FY2024 (32.7%) had criminal histories and highlighting prosecutions and thousands of criminal indictments tied to its arrests [1] [4] [5]. ICE’s materials also distinguish administrative arrests—defined as detaining someone the agency believes is removable—from criminal arrests and note that ERO historically arrests people with convictions ranging from DUI and drug possession to assault and certain traffic offenses [5] [4]. Those official tallies serve as the baseline the agency uses to defend its prioritization of “criminal” removals [1].
2. Think tanks and the media documenting a rising share with no convictions
Independent analysts and several media outlets report a markedly different composition: some studies find the majority of interior arrests in recent periods involved people without criminal records. For example, a Cato Institute analysis cited by the Brennan Center estimated that between Oct. 1, 2024 and June 16, 2025 ICE arrested roughly 204,000 people and that about 65 percent—approximately 133,000—had no criminal record [2]. Stateline’s reporting, using the Deportation Data Project and university research, similarly flagged a rising share of arrests solely for immigration violations—rising from about 20 percent in April to 44 percent in October of one recent year—and found the Biden-era share earlier in 2024 was around 10–11 percent under a different metric [6].
3. Methodological fault lines that explain conflicting portraits
The divergent conclusions trace to methodological choices: ICE’s internal reporting separates administrative and criminal arrests and counts removals by whether an individual has a recorded criminal history [1] [5], while external projects such as the Deportation Data Project rely on FOIA-released, linked datasets that let researchers trace arrest pathways but carry documented quality issues and gaps [7] [8]. Migration Policy and USAFacts emphasize that “administrative” arrests can include people with past convictions as well as those without, and that in FY2024 roughly half of ICE’s administrative arrests were of people with no recorded convictions [3] [9]. Reporting platforms also differ on whether they include HSI criminal arrests, transfers from CBP after border apprehensions, or only interior ERO actions—choices that materially alter the criminal vs. noncriminal shares [9] [10].
4. What changes in enforcement strategy have journalists and analysts highlighted
Media and think-tank work points to a wider shift in resources from interior removals toward border processing and expedited mechanisms, producing two parallel trends: border removals surged (224,000 in FY2024) while removals from the interior fell over the past decade, and interior enforcement increasingly produced arrests for immigration status rather than new serious-crime prosecutions [10] [3]. Analysts also link arrest concentrations to state and local cooperation: jurisdictions that require local-federal collaboration report higher ICE arrest rates, suggesting placement and policy choices amplify interior enforcement’s noncriminal component [11].
5. Bottom line: consensus, caveats, and where reporting diverges
The emerging consensus among journalists and think tanks is that a larger share of ICE’s enforcement in and since 2024 consists of people without recent criminal convictions than agency messaging implies, but the precise share varies widely by dataset and definition—ICE’s FY2024 headline that 32.7% of removals had criminal histories [1] sits alongside independent estimates that noncriminal cases constitute a majority in some recent windows [2] [6]. Crucially, analysts caution that FOIA-derived datasets and ICE dashboards each have limits; careful comparison requires harmonizing definitions (administrative vs. criminal arrest, interior vs. border removals) and acknowledging gaps and duplicates flagged by the Deportation Data Project [7] [8].