The traced-gun total and the border seizures measure different things
GAO reported that 82,785 firearms recovered in Mexico and traced from January 2020 through December 2024—68% of the traced total—had a U.S. source. CBP separately seized 4,944 firearms headed to Mexico at land ports from January 2020 through December 2025.
Those numbers cannot be subtracted to estimate a seizure rate. The trace population includes guns recovered in Mexico and submitted for tracing; the CBP count covers interdictions at land ports, across a different six-year period. Neither measures every gun trafficked.
| Measure | Count | Period |
|---|---|---|
| Recovered in Mexico, traced to U.S. source | 82,785 (68%) | 2020–2024 |
| CBP Mexico-bound land-port seizures | 4,944 | 2020–2025 |
| ATF/State assistance reviewed | $126 million | FY 2020–2025 |
Agencies counted activity without measuring effectiveness
GAO found that DHS, ATF and the State Department reported outputs such as seizures and training, but had not all established goals, baselines and performance measures needed to judge effectiveness. CBP data also did not consistently identify the semi-automatic rifle types sought by trafficking organizations.
The accountability finding is therefore not that the programs failed. It is that agencies spending money and deploying inspectors could not connect activity counts to defined results.
One recommendation drew disagreement
GAO issued four recommendations: analyzable firearm-type data, DHS objectives and outcomes, ATF measures for Operation Southbound, and State baselines and targets. Agencies agreed with three; DHS disagreed with the firearm-data recommendation. GAO maintained it was valid. All four were open on September 28.
The report also attributes the trafficking chiefly to straw purchasers and middlemen. That is an agency assessment of trafficking patterns, not an allegation that every traced firearm was knowingly sold for a cartel.
Method and limitation
The newsroom compared GAO’s ATF trace, CBP seizure and funding figures, checked periods and denominators, and reviewed each recommendation’s status. GAO itself analyzed agency data and interviewed U.S. and Mexican officials.
Trace data are not a random sample of all Mexican crime guns, and a U.S. source does not by itself identify the purchaser, smuggler or lawful chain of possession.