How do ICE and independent groups define and count 'criminal convictions' versus 'pending charges' in detention statistics?
Executive summary
ICE’s public detention statistics sort detainees into at least three explicit administrative buckets — those with criminal convictions, those with pending criminal charges, and those with neither — but federal officials and political appointees sometimes lump convictions and pending charges together when describing “criminal” detainees, a choice that independent analysts and fact-checkers say makes the detained population look more criminalized than separate counts indicate [1] [2] [3].
1. How ICE defines the categories in its dataset
ICE’s official materials and dataset categorize detained individuals as having criminal convictions, having pending criminal charges, or having neither conviction nor pending criminal charge; the “neither” group also includes immigration-specific cases such as reentry-after-deportation, immigration fugitives with final removal orders, and foreign fugitives wanted abroad [1].
2. What “pending charges” means operationally and in ICE programs
ICE and its enforcement partners treat “pending” or “active” criminal charges as a distinct administrative status that can be used to prioritize custody and transfers — for example, the 287(g) jail enforcement model explicitly identifies and processes removable aliens who have pending or active criminal charges while in local custody — but those pending charges are, by definition, not convictions and may later be dismissed or downgraded [4] [1].
3. How the agency and political officials report the counts — and why that matters
Senior DHS and ICE officials have sometimes described detainees with either convictions or pending charges as “criminal illegal aliens,” combining the two categories in public statements to emphasize public‑safety goals; critics and fact‑checkers note that this conflation inflates the appearance of criminality because pending charges are not convictions [2] [3].
4. How independent analysts and reporters treat the two categories
Think tanks, journalists and fact‑checkers generally report convictions and pending charges separately and point out that a large share of people in ICE custody have no criminal conviction: multiple analyses found roughly 70–72% of people in detention lacked a conviction at snapshot points in 2025, with about one quarter of the detained population having pending charges and roughly 47% having neither convictions nor pending charges in one ICE download examined by Snopes and others [2] [3] [5].
5. What the numbers say about violent convictions and the limits of the datasets
Independent breakdowns indicate only a small slice of detainees have violent convictions — one analysis put violent convictions at about 5% of detainees — and the public ICE arrest datasets have important limits: they may omit Border Patrol referrals in some views, do not always disclose charge severity or jurisdictional nuances (e.g., foreign convictions or Interpol notices), and can mix people arrested by ICE with those transferred from other agencies, all of which complicates headline percentages [6] [7] [8].
6. Competing narratives and implicit agendas in counting choices
Counting pending charges with convictions serves an enforcement-centric narrative that underscores the administration’s “worst of the worst” rhetoric and justifies aggressive operations, while independent researchers emphasize due process and the difference between allegations and adjudicated guilt; both choices are defensible tactically but they produce very different public impressions of who ICE detains [2] [6] [8].
7. Practical takeaway for readers interpreting ICE statistics
When reading ICE numbers, treat “convictions” and “pending charges” as distinct metrics: ICE’s raw data provides both categories, but public statements may conflate them; trustworthy independent analyses separate the two, highlight the share with neither charge nor conviction, and flag dataset omissions so that policymakers and the public can judge enforcement claims against the underlying case‑level reality [1] [3] [5].