How does ICE define 'criminal' in its arrest statistics and how do pending charges figure into that count?
Executive summary
ICE’s public tallies of “criminal” arrests mix three different things: (a) people with criminal convictions, (b) people with pending criminal charges, and (c) enforcement arrests for immigration violations that are administrative (civil) in nature; the agency and DHS often present the combined figure as a single “criminal” category, a practice critics say inflates the share of truly convicted offenders (ICE’s own categories and DHS statements) [1] [2]. Independent analyses and watchdogs find a much larger share of detainees with no criminal convictions, arguing that treating pending charges or immigration-only violations as equivalent to convictions skews the public picture and masks policy choices [3] [4] [5].
1. What ICE’s reports actually label as “criminal” — mixed categories, not just convictions
ICE’s public statistics and DHS press releases use language that groups together people “charged or convicted” or otherwise described as “criminal illegal aliens,” but ICE’s underlying categories include those with convictions, those with pending charges, and a broader third bucket—people with no convictions or charges who are detained for immigration violations such as re‑entry after removal or having a final order of removal (ICE’s statistics pages and DHS releases make these distinctions even as headlines do not) [1] [2].
2. How pending charges are counted and why that matters
DHS and ICE frequently report the share of arrests that are of individuals “charged or convicted,” and department spokespeople have cited that combined figure (for example, a repeated DHS claim that roughly 70% were “charged or convicted”) [2] [6] [7]. Policy analysts and researchers warn this conflation is consequential because pending charges are allegations, not findings of guilt; Cato’s analysis flagged that DHS/ICE count people with pending criminal charges as “criminal arrests,” even when charges are minor or later dismissed, and that practice inflates the impression that ICE’s operations are primarily removing convicted violent offenders [3] [8].
3. Independent data paint a different distribution of convictions versus no-record cases
Multiple independent trackers and analyses report that a large share of people in ICE custody lack criminal convictions: TRAC reported roughly 73.6% of people held in ICE detention had no criminal conviction as of late November 2025 (48,377 of 65,735) [4], and Cato/Brennan Center analyses found that roughly 65–73% of those arrested or detained had no criminal record, with only a small fraction convicted of violent offenses [3] [5]. These findings underscore that counting pending charges alongside convictions materially changes the headline percentages cited by DHS [3] [5].
4. Administrative arrests vs. criminal arrests: legal and reporting distinctions
Journalistic and legal guidance stresses that many ICE arrests are administrative—civil arrests for immigration violations—not criminal arrests, and ICE’s ERO data focuses on administrative arrests while other ICE components (HSI) may conduct criminal investigations that are separate and not always reflected in the same datasets (a clear distinction explained in reporting that unpacks ICE arrest data) [9]. The statutory rules that allow ICE officers to make criminal arrests in limited circumstances (e.g., felonies observed in their presence or with exigent risk) mean some criminal arrests occur, but that does not justify treating all non‑convicted suspects as convicted criminals in headline statistics [9].
5. Why the dispute persists — incentives, narratives, and limits of the public data
DHS and ICE have an explicit enforcement narrative—“worst of the worst”—and present aggregated metrics (charged or convicted) that bolster it, while advocacy groups and researchers emphasize granular conviction data and the rise in detentions of people with no convictions to challenge that narrative [2] [10] [3]. Public datasets are imperfect: ICE releases summary categories but not always the underlying charge dispositions or differentiations among types of charges, and independent analyses rely on leaked or compiled datasets to show how pending charges and immigration arrests change the picture; therefore conclusions depend on which categories and timeframes one examines [1] [3] [4].