997 Recalls Covered 29 Million Cars Last Year. Almost None Were the Ones Killing People.

I ran a cross-tab that should not have surprised me and did anyway. NHTSA issued 997 safety recalls covering more than 29 million vehicles in 2025.^[1] Traffic deaths fell 6.7% to an estimated 36,640, the second-lowest fatality rate in recorded history at 1.10 per 100 million miles traveled.^[2] Correlation? I pulled the recalled models against FARS fatality rates. Overlap is almost nonexistent.

FARS tracks 337 vehicle models with 50 or more deaths over the 2014-2023 period. Rank them by death rate per 100 million VMT. In the top quartile, 84 models with rates above 1.15 account for 62.9% of all occupant deaths while representing 29.2% of the fleet.^[3] In the safest quartile, 84 models with rates at or below 0.26 hold 22.1% of the fleet and produce 3.4% of deaths. Eighteen times the death share relative to fleet share, top to bottom.

Now look at 2025's biggest recalls. Ram 1500's seatbelt anchor campaign covered 1.27 million trucks at a FARS rate of 0.78. Jeep's Wrangler fire recall hit 1.08 million vehicles at 0.84. Hyundai's Tucson phantom braking recall reached 421,000 at 0.34. Tesla's Model 3 and Model Y suspension probe covered 1.2 million vehicles with rates of 0.05 and 0.03, respectively.^[4] Every single one sits in the safer half of the distribution. Only the Explorer seat recall at 1.54 cracks the top quartile's threshold among major 2025-2026 campaigns.

Recalls are not broken. They do exactly what they were built to do: identify manufacturing defects, notify owners, issue fixes. A seatbelt anchor that was never bolted down is a defect. An engine that seizes at 70,000 miles is a defect. A steering pump that catches fire while parked is a defect. All of these deserve campaigns. But defects are not why 36,640 people died last year. Models with the highest FARS death rates are not there because of a bolt or a sensor. They are old. They lack electronic stability control, automatic emergency braking, modern crumple zones, and curtain airbags. A Chevrolet Cavalier (rate: 2.43, 1,225 deaths), Dodge Neon (1.50, 602 deaths), or Buick LeSabre (2.67, 1,344 deaths) will never receive a recall because there is nothing to fix. Design is the defect.

What this means if you are buying a car: check your VIN for open recalls at nhtsa.gov/recalls. Fix whatever comes up. Then understand that completing a recall on a 2006 Impala does not make it a safe car. Your single largest lever for reducing fatal crash risk is driving something built after 2015, when ESC, side curtain airbags, and rearview cameras became mandatory. Every model year you step forward reduces your fatality risk. No recall campaign delivers that.

What this analysis does not prove: Recalls prevent injuries and deaths from specific manufacturing defects regardless of a model's FARS rate. A seatbelt anchor failure kills the same way in a Tesla as in a Cavalier. This cross-tab cannot measure how many lives individual recalls saved. It only measures where the system's aggregate tonnage lands relative to where the deaths concentrate, and the answer is: mostly somewhere else.