The Chevy S-10 Died in 2004. It Has Killed 1,427 People Since.
General Motors pulled the plug on the Chevrolet S-10 in 2004, replacing it with the Colorado. A reasonable corporate decision: the compact pickup market was shrinking, the platform was old, and GM needed the factory floor for something newer. What GM did not do was pull the trucks off the road. Between 2014 and 2023, those orphaned S-10s killed 1,427 people.[1]
That rate puts the S-10 among the five deadliest vehicles in the entire FARS database, outpacing the Ford Mustang, the Nissan Maxima, and the Chevrolet Impala. It is 37 times the fatality rate of a modern Ram 1500. And the impairment data offers no alibi: 20.5% of S-10 fatal-crash drivers tested positive for alcohol or drugs, which is dead-average for the national fleet.[1] Sober people are dying in this truck at exactly the rate you would expect from a vehicle that predates every meaningful crash-avoidance technology on the market.
The S-10 was built on a body-on-frame platform with rear-wheel drive and a chronically light rear axle when running empty, which was most of the time since few compact-truck owners haul near capacity. That empty bed makes the rear end a pendulum in rain, snow, or an emergency lane change. The truck shipped without electronic stability control because FMVSS 126 did not require it until model year 2012, eight years after GM shut the line down.[2] No amount of recalls can fix this deficit. ESC is not a bolt-on accessory; it requires a vehicle architecture designed around it from the braking hardware up through the steering geometry. Every S-10 still registered in the United States is permanently locked out of the single technology that IIHS credits with cutting fatal single-vehicle crashes by roughly 56%.[3]
GM also sold the same truck as the GMC Sonoma. Add the Sonoma's 378 deaths and the platform's decade-long toll rises to 1,805. That is more people than the Porsche 911, BMW Z3, Mazda MX-5 Miata, Nissan 370Z, and Tesla Model Y have killed combined.[1] The S-10/Sonoma body count exceeds the entire FARS death total of several current-production models that outsell it ten to one. These trucks are ghosts in the fleet, invisible on new-car safety brochures, absent from NHTSA's 2026 test slate, yet showing up in the fatality ledger at a rate that would trigger an emergency investigation if they were still being manufactured.
When an S-10 does crash fatally, 75.5% of those crashes produce a death. That crash-lethality figure sits among the worst in any vehicle class, not just pickups. For context, the class average for pickups with fleets above 50,000 is around 49%.[1] The S-10's number tells you something precise about the structural collapse pattern: the cab crumples inward rather than distributing force outward, and there are no side curtain airbags to catch what the structure misses.
NHTSA announced this month that America's Q1 2026 fatality rate reached 0.99 per 100 million VMT, tied for the second-lowest quarterly figure in the nation's history.[4] For the 218,750 S-10s still estimated to be in the fleet, that headline describes somebody else's country. Their occupants drive on roads that are functionally five times more lethal than average, in a truck whose manufacturer no longer exists to issue recalls, built to standards that the federal government abandoned more than a decade ago. The fix is the same one NHTSA will never mandate for a vehicle it no longer regulates: stop driving it.
Sources & References
- NHTSA, Fatality Analysis Reporting System (FARS), 2014–2023. nhtsa.gov
- NHTSA, FMVSS No. 126: Electronic Stability Control Systems, Final Rule, 72 FR 17236, April 6, 2007. govinfo.gov
- IIHS, “Life-saving benefits of ESC continue to accrue,” 2011. iihs.org
- NHTSA, Early Estimates of Motor Vehicle Traffic Fatalities, January–March 2026, DOT HS 813 829, July 2026. nhtsa.gov
Source: NHTSA FARS 2014–2023. Fleet estimates are derived from registration-weighted VMT projections and carry ±15% uncertainty for low-volume models. The S-10 fleet estimate (218,750) may overstate or understate the actual registered population. See methodology for caveats.