How did ICE change its removals-counting methodology in July 2023 and what effect did that have on historical totals?

Checked on January 11, 2026
Disclaimer: Factually can make mistakes. Please verify important information or breaking news. Learn more.

Executive summary

ICE adopted a new internal rule in July 2023 that changed which movements it classifies as “ICE Removals,” applied that rule retroactively to May 12, 2023, and as a result modestly increased the agency’s published removal totals for that window while creating confusion across multiple public datasets and third‑party trackers [1] [2]. Independent trackers and ICE’s own releases diverged until ICE clarified which event types were to be counted under the new methodology, prompting corrected data files and adjustments to historical tallies [1] [2] [3].

1. What the methodological change actually did

The policy shift broadened the operational definition of ICE Removals to include noncitizens processed under Expedited Removal (ER) or recorded as Voluntary Return (VR) who were subsequently turned over to ICE’s Enforcement and Removal Operations (ERO) for detention and removal, and — as of May 12, 2023 — those ER cases handed from Customs and Border Protection (CBP) to ICE for removal via ICE Air [1]. ICE’s own public data products distinguish ICE removals from CBP removals and Title 42 expulsions, so the change was about which intra‑DHS handoffs and procedural categories count toward the ICE Removals line in ICE’s reporting [3] [4].

2. How the change was applied in practice

ICE implemented the counting rule in July 2023 but applied it retroactively to records back to May 12, 2023 — the date Title 42 expulsions ended and Title 8 procedures resumed at the border — meaning datasets covering mid‑May onward were subject to restatement when ICE adopted the new standard [1]. The Deportation Data Project says the July methodology was applied retroactively and that initial ICE data files (a “late June” release) were incorrect and later removed, while a corrected “late July” file reflected the updated counting convention and was judged complete by the project [1] [2].

3. Measurable effect on historical totals

The immediate numeric impact reported by third‑party analysts was measurable but not astronomical: the Deportation Data Project noted that the corrected late‑July removals file recorded 282,214 removals for FY2024 while ICE’s 2024 annual report lists 271,484 — a difference of roughly 10,730 attributable at least in part to the differing counting conventions and later data corrections [1] [5]. In short, the methodological change and its retroactive application increased ICE’s removals count for the affected period and produced a revisionary bump in published totals once ICE clarified which events it would include [1].

4. Why data users and watchdogs raised alarms

Watchdogs and data aggregators flagged the change because ICE’s reporting has a documented history of opaque definitions and shifting coverage; the Government Accountability Office previously criticized ICE for understating detention counts and recommended stronger data reporting practices, which amplifies the consequences when ICE alters counting rules midstream [6]. The Deportation Data Project and others warned users not to rely on the earlier, incomplete removals table and urged caution until the corrected files were posted [2] [7].

5. Broader context and remaining uncertainties

Counting rules for “removals” versus “returns,” and how to treat CBP expulsions under Title 42 or other agencies’ actions, have long complicated historical tallies — official yearbooks and DHS notes show prior counting revisions (e.g., changes in administrative‑arrest counting) and long‑standing distinctions between removals and returns [8] [9]. Available sources document the policy change’s scope and its retroactive application but do not publish a line‑by‑line algorithm showing exactly which record fields were toggled; therefore some reconciliation across ICE dashboards, the annual report, and FOIA releases required interpretive work by third‑party projects [1] [2] [3].

6. Bottom line for researchers and reporters

The July 2023 methodological change caused ICE to count additional categories of cases as “ICE Removals” beginning May 12, 2023, producing a retroactive uptick in published removal totals for that window and triggering corrected data releases and public disputes between ICE and independent data projects; the net numeric effect on FY2024 totals is on the order of tens of thousands in the datasets highlighted by Deportation Data Project, but precise reconciliations depend on which ICE files (original, late‑June erroneous, or late‑July corrected) are used [1] [2] [5]. Sources do not provide a complete, public mapping of every data field changed, so analysts must continue to cross‑check ICE’s dashboard, annual reports, and vetted FOIA‑derived files [3] [7].

Want to dive deeper?
How do ICE, CBP, and DHS differ in their definitions of removals, returns, and expulsions?
What specific record fields and event types changed in ICE’s July 2023 removals dataset, according to FOIA releases?
How have third‑party trackers (Deportation Data Project, TRAC, ACLU) reconciled ICE removal counts across successive data releases?