What percentage of those detained by ice do not have criminal records
Executive summary
Recent reporting and data analyses converge on a clear but variable finding: a substantial share—and in many snapshots a majority—of people in ICE custody lack a recorded U.S. criminal conviction, but the exact percentage depends on the dataset, the date, and whether one measures “arrests” versus the current “detained population” or counts convictions versus pending charges; mainstream analyses place that share roughly between the low‑20s in certain historical arrest windows and as high as the mid‑70s among detainees in late 2025 [1] [2] [3].
1. What the headline numbers say — conflicting snapshots
Multiple independent analyses of ICE data and leaked datasets report different percentages because they are looking at different slices: TRAC and related trackers reported that about 73.6% of people in ICE detention had no criminal conviction as of November 30, 2025 (48,377 of 65,735) [2], the Cato analysis placed roughly 69% of current detainees without convictions by mid‑November 2025 [3], The World Data’s summary for January 16, 2026 estimated about 53% of detainees had no U.S. criminal record [4], and FactCheck’s look at arrest records found 21.9% of those arrested in the first three months of Trump’s term had no U.S. criminal record — illustrating how arrest-era snapshots can differ from the composition of the detained population [1].
2. Why the percentages vary — definitions and data slices matter
Part of the divergence is definitional: ICE distinguishes people with convictions, people with pending charges, and those with no convictions or pending charges but who may be immigration violators, re‑entrants, fugitives, or wanted abroad [5]; datasets also mix arrests (who ICE charges), transfers from CBP, and the “detainee census” at a given date, and some analyses exclude Border Patrol referrals or only cover arrests captured in a leaked Deportation Data Project file — thus a figure for “arrests with no criminal record” is not directly comparable to “percentage of the detained census without convictions” [1] [5] [6].
3. Temporal dynamics — big recent shifts reported by researchers
Multiple organizations documented a dramatic rise in detentions of non‑convicted individuals over 2024–2025: TRAC and the American Immigration Council reported huge percentage increases in the number of people with no criminal history held in ICE custody — including claims of thousands more detainees without convictions and growth that accounted for the bulk of detention increases [7] [8] [9]. University and think‑tank analyses also found monthly detentions of Latinos without criminal records increased sharply compared with the Biden era [10].
4. How to answer the user's question directly — a reasoned range, not a single static number
Given the different sources and timeframes, the most defensible direct answer is a range: recent reporting from late 2025 into January 2026 places the share of people in ICE custody who do not have a recorded U.S. criminal conviction between roughly 50% and 74% in major independent counts [4] [2] [3]. Historical snapshots of arrests can be lower (FactCheck noted 21.9% for arrests in a specific early‑term window) because arrests and the detained census capture different populations and periods [1]. Readers should treat any single percentage as a snapshot tied to a data definition and date [5].
5. Caveats, alternative views and institutional context
ICE’s public statistics and its categorical definitions complicate interpretation — the agency notes people with “no convictions or pending charges” can include immigration re‑entrants or foreign fugitives, and custody counts are driven in part by transfers from CBP at the border [5]; advocates and academic centers emphasize the policy choice behind who is targeted and detained, arguing increases in non‑convicted detainees reflect shifts in enforcement tactics [8] [10], while other summaries stress that arrest datasets versus detained‑population snapshots will yield different percentages [6] [1]. Where possible, rely on the specific dataset and date cited when a percentage is quoted.