Where can I download ICE’s official FY2025 and FY2026 arrest datasets (CSV) and what fields do they contain?

1. Where the CSVs live: ICE’s official portal and the Deportation Data Project mirror

ICE publishes enforcement statistics and periodic data releases on its official statistics and enforcement dashboard pages; those are the primary “official” sources for quarterly and fiscal-year data ^{[1] [6]}. Independent groups obtained ICE’s machine-readable tables via public records and litigation and now host them in a more researcher-friendly index at the Deportation Data Project, which posts the actual CSV/XLSX files, bundles by release date, and tracks successive ICE releases ^{[2] [4]}. For practical downloading, the Deportation Data Project’s data page is the most direct place to fetch arrest tables for FY2025 and FY2026; ICE’s own statistics pages provide the official source and dashboard visualizations that point to underlying releases ^{[2] [1]}.

2. Which specific files correspond to “arrests” and where to click

The datasets labeled as “arrests” or “administrative arrests” appear in the Deportation Data Project’s ICE data collection alongside related tables for detainers, detentions, encounters, and removals; the project’s documentation page describes which table is the arrests table and links to the downloadable files ^{[3] [2]}. ICE’s enforcement dashboard and statistics pages offer quarterly and fiscal-year snapshots that map onto those same administrative tables, so cross-checking the Deportation Data Project file timestamps with ICE’s dashboard releases ensures the researcher is using the correct FY2025 or FY2026 CSV ^{[1] [7]}.

3. What fields the arrest CSVs contain (and how reliable the field list is)

The Deportation Data Project provides a field-level description for ICE’s administrative tables — including the arrests table — so the authoritative list of variables for the arrests CSV is in that documentation rather than a secondary article; it covers demographic fields, timestamps, enforcement action codes, processing location, and related administrative identifiers ^[3]. ICE’s own data collection practice is to record every person it encounters, arrests, detains, transports, and deports, implying the arrest table contains encounter/booking dates, place (facility/jurisdiction), enforcement source (e.g., ICE-initiated, transferred from state/local jail), and person-level demographics and administrative IDs — researchers should consult the Deportation Data Project’s variable glossary to see each exact column name and codebook ^{[2] [3]}.

4. Known data-quality caveats and recent release history

Independent curation matters: the Deportation Data Project not only republishes files but also issues warnings when ICE releases contain anomalies, recently recommending specific late-July release files over earlier June exports because of missing or misclassified observations in removals — a reminder that arrests and related tables have had documented inconsistencies across releases and dates ^[5]. The Deportation Data Project’s tracker and release notes (and ICE’s quarterly dashboard updates) should be consulted to pick the most accurate FY2025/FY2026 CSV and to understand provenance for any suspect spikes or gaps ^{[7] [5]}.

5. How to proceed with downloads and verification

Best practice is to download the arrests CSV from Deportation Data Project’s indexed data page (which links original ICE exports and provides documentation), then cross-check the file timestamp and row counts against ICE’s official statistics/dashboard release for the matching quarter or fiscal year to confirm provenance and completeness ^{[4] [1]}. When analyzing fields, use the Deportation Data Project’s documentation for column definitions and the project’s advisories about which release versions are recommended, because they explicitly remove or flag problematic earlier files ^{[3] [5]}.