Colombia's mining agency said accidents fell 38.5%. Deaths did not.

In April the same agency announced that mining accidents had fallen 38.5%. It compared accidents, not deaths, in the first quarter: 36 in 2025 and 22 in 2026 (by our arithmetic, a 38.9% fall). We checked that against the mine-rescue register we obtained by petition. The fall in accidents is real. Deaths did not fall.

In short: Colombia has fewer mine emergencies this year and more people dying in them. Through 24 September, emergencies are down 24% and deaths are up 24%. The April release gave no death count.

Most emergencies never reach ANM's news page. From 1 January to 24 September the register holds 66 emergencies with at least one death. ANM's news page carried a release on 7 of them, which account for 47 of the 124 deaths. None of the 46 single-death emergencies got a release there. Neither did three larger ones: Diamante 42 in Socha on 16 January (4 dead), a gold working in Remedios on 3 April (3 dead), and La Quinta in Cucunubá on 9 May (4 dead). Colombian newspapers reported Socha at the time. We read ANM's news page, not its social accounts or its statements to reporters.

One emergency reads differently in the two records. On 5 June ANM said one person had been found dead and six were trapped at a coal working in Sutatausa that it had not authorized, and it called the working illicit. We found no later release on that emergency. The register has 7 people involved and 7 dead on 4 June at the same vereda, filed as legal, against San Antonio 1 y 3. ANM has changed legal readings in its records before, so this may be a later correction. We do not know which reading is right.

Most deaths this year happened at legal operations. At 24 September the register held 71 deaths at legal workings and 53 at illicit ones. ANM's split on 21 October counts emergencies only: 57 legal and 31 unauthorized.

Nariño is the new part. The register holds 14 deaths in Nariño from 2012 through 2025. This year two gold collapses there, two months apart, killed 27: 13 in Policarpa on 19 August and 14 in Iscuandé on 19 October. ANM's Iscuandé statement calls it a public order problem and says responsibility falls on the territorial authorities, not on the agency. The register reaches us by petition, so Iscuandé will not appear in it until the next delivery.

Between the register's last row and ANM's 21 October count, the totals rose by 12 emergencies and 24 deaths. ANM's news page named two of those emergencies, Timbiquí (3 dead) and Iscuandé (14). That difference assumes the statement and the register count the same way. In earlier years the register matched ANM's published counts, but we cannot check this year's statement row by row.

For anyone who wants the detail, here is how we did it. The register is ANM's answer to our public-records requests. Its newest row is 24 September 2026, and it covers only emergencies reported to ANM and handled by its rescue service, not every mining death in Colombia. ANM revises these figures. We read every release on ANM's news page for 2026 and matched each one by hand to a register row by date and municipality. The register is on /colombia and in Spanish at /es/colombia.

ANM's April release: www.anm.gov.co/anm-reduce-en-385-accidentes-mineros-gracias-inversiones-oportunas-capacitaciones-y-articulacion

ANM's Iscuandé statement: www.anm.gov.co/anm-lamenta-la-muerte-de-14-personas-en-narino-y-llama-reforzar-la-seguridad-minera