What major investigative reports have documented ICE use-of-force incidents since 2009 and how did they count shootings?

1. Major investigations that produced public counts

The Trace published a long-form investigation cataloguing multiple ICE shootings and how the agency handled use-of-force reporting and policy changes, documenting incidents including apparently unarmed victims and inconsistencies in reporting requirements ^[1]. Business Insider, working with Type Investigations, obtained hundreds of pages of ICE training and internal materials and published a complementary report and database that found dozens of incidents and identified systemic gaps in training and reporting ^{[4] [5]}. Legal and advocacy summaries and secondary reporting — for example Criminal Legal News’ synthesis and national outlets like NBC and the BBC — have amplified those tallies and added contemporaneous incidents from late 2024 into 2026 to their running lists ^{[3] [6] [7]}.

2. What their headline counts look like

Business Insider/Type-linked reporting and aggregated records have been cited as finding at least 59 shootings between 2015 and 2021, with roughly 23 people killed and 24 injured during that stretch — figures repeated in legal and advocacy summaries ^[3]. The Trace’s reporting documented a list of shootings, including a pattern of shootings in traffic settings and public places and instances where victims appeared unarmed; it also tied reporting back to a narrower DHS data request that only required agencies to supply use-of-force data beginning in 2018, which limits direct comparability ^[1].

3. How investigators counted “shootings” — inclusion rules and sources

Investigative teams counted incidents as shootings when there was documentary or corroborating evidence that an ICE agent (or a federal immigration enforcement operation involving ICE) discharged a firearm; evidence included ICE incident reports, body-camera footage, medical and coroner records, local police reports, civil lawsuits and depositions, and where available, contemporaneous video ^{[1] [2]}. Reporters supplemented incomplete FOIA returns by suing for records or using leaked/training documents and public filings to build incident lists; Business Insider explicitly used hundreds of pages of training and incident material to identify cases ^{[2] [4]}.

4. Time windows, DHS policy changes and counting complications

DHS’s 2021 handbook changed reporting timing and scope — for example altering when agents must “immediately” report an incident versus within an hour if safe — and the department’s 2022 directive required agencies to report use-of-force data but asked only for data back to 2018, creating a discontinuity for journalists trying to assemble a longer historical list ^{[1] [2]}. Journalists therefore combined older case-by-case sources (lawsuits, local files) with the newer DHS-collected data, resulting in different headline totals depending on whether a project started counting in 2009, 2015, or 2018 ^{[1] [3] [2]}.

5. Sources of undercounting and why tallies diverge

ICE’s historical opacity — frequent reliance on FOIA exemptions, outsourcing of investigations to local agencies whose records are incomplete, and refusal to publish full use-of-force policies or incident lists — means public counts likely understate total incidents and make independent verification difficult; reporters explicitly describe being “stymied” by withheld records and state exemptions that blocked access to incident reports ^{[3] [2]}. Contemporary outlet tallies (NBC, BBC) that add recent shootings since September 2025 show how dynamic the record is and how differing inclusion choices (e.g., Border Patrol vs. ICE, federal task force operations) change totals ^{[6] [7]}.

6. Bottom line: what can confidently be said

The best-documented, peer-reviewed journalistic counts since roughly 2009 come from The Trace and Business Insider/Type Investigations, which used public records, litigation, and internal documents to compile incident lists; those projects reported dozens of shootings, with one synthesis noting 59 shootings from 2015–2021 and dozens killed or injured, but all investigators warn their counts are conservatively limited by withheld records and DHS’s own reporting window that only reliably begins in 2018 ^{[1] [2] [3]}. Alternative counts from mainstream outlets and government statements add more recent incidents and emphasize that methodology and scope — not mere arithmetic — drive the differences between lists ^{[6] [7]}.