Where can researchers download ICE’s biweekly detention and removals tables for 2025–2026 and how are fields coded?
Executive summary
Researchers seeking ICE’s biweekly detention and removals tables for 2025–2026 will primarily find them on ICE’s public statistics page, with parallel copies and processed versions hosted by academic and advocacy projects such as the Deportation Data Project and archival efforts cited by the Guardian and Vera Institute; ICE warns that these files “fluctuate until ‘locked’” at fiscal year close [1][2][3]. Field-level documentation and a codebook that explains variable definitions, row structure, and the logic used to collapse or flag duplicates are available from the Deportation Data Project, which also documents known dataset irregularities and recommends particular snapshots for removals [4][5][3].
1. Where to download the biweekly tables (official source and mirrors)
The authoritative public download location is ICE’s Enforcement and Removal Operations statistics page, which publishes arrest, removal, detention and alternatives-to-detention statistics — including the biweekly detention spreadsheets — and states that it cannot attest to subsequent transmissions because data “fluctuate until ‘locked’” at fiscal-year end [1]. DHS’s Office of Homeland Security Statistics (OHSS) also publishes monthly immigration enforcement tables that draw from the same operational Persist Dataset and can provide complementary monthly tallies [6]. Independent archival and analytic projects mirror and republish ICE’s biweekly releases: the Deportation Data Project hosts original and processed ICE datasets and interactive tools (and links to specific late‑July removals snapshots it recommends) while news organizations like The Guardian have been archiving each biweekly release since January 2025 [3][2][5].
2. How the files are structured and how fields are coded (row/variable logic)
ICE’s individual‑level enforcement tables (arrests, detainers, encounters, removals) are structured so that each row represents a single enforcement action for a particular noncitizen, meaning individuals may appear multiple times if arrested, detained, or removed more than once; the detentions table is more complex because a single stay can generate multiple facility-level rows [4]. The Deportation Data Project’s codebook and documentation explain field names, value coding, and the distinction between dataset types, and they provide processed “stay-level” detentions data that collapse multiple facility transfers into one stay record [4][7]. ICE’s biweekly detention spreadsheets also include fiscal‑year‑to‑date counters for effectuated removals and categorical fields for citizenship, criminality, arresting agency, facility, and custody status; the OHSS monthly tables similarly include standardized fields for book‑ins, book‑outs, removals, returns and parole drawn from the Persist Dataset [8][6].
3. Important coding changes, omissions and recommended snapshots
ICE revised its removals counting methodology in July 2023 and applied it retroactively (affecting how expedited removals and voluntary returns are counted when turned over to ERO), which means researchers must account for that rule change when comparing across time [5]. The Deportation Data Project has flagged that certain ICE removals releases (notably early June and late June 2025 files) were incomplete or incorrect and has recommended using the late‑July removals dataset for analytic consistency; researchers should consult those notes and prefer archived snapshots identified by the project [5][3]. ICE and OHSS footnotes and dataset “sourcing notes” are essential reading because ICE explicitly cautions that posted files may be revised until the fiscal year is locked [1][6].
4. Practical tips for researchers working with the tables
Download the original ICE biweekly XLSX/CSV from the ICE statistics page and also fetch the matching processed versions and codebook from the Deportation Data Project to map variable names and value labels; use the Deportation Data Project’s “stay‑level” detentions table if the research question requires one‑row‑per‑detention analyses [1][4][7]. Cross‑check biweekly files against OHSS monthly tables and archived biweekly snapshots (Guardian/Vera archives or the Deportation Data Project) to detect missing removals or mid‑release corrections, and explicitly document which ICE snapshot and which codebook version were used because ICE’s removals totals and counting rules have changed and files “fluctuate” [6][5][2].
5. Caveats, competing narratives and what reporting sometimes misses
Public reporting often cites headline removal totals from the biweekly spreadsheet without noting ICE’s methodological change in 2023 or the fact that some published removals files have been incomplete and later corrected — points the Deportation Data Project and GAO‑style audits emphasize [5][9]. OHSS monthly tables offer complementary perspectives but draw from DHS’s Persist Dataset and are updated on a different cadence, so discrepancies across ICE biweeklies, OHSS monthlies, and third‑party processed datasets are predictable and must be reconciled rather than assumed to be errors [6][3].