The Average Car That Kills You Is 12.5 Years Old

I ran every model-year death count in FARS across 323 tracked vehicle models, weighted each fatality by approximate vehicle age at the midpoint of the 2014–2023 data window, and got a number nobody seems to have published before: 12.5 years.^[1] That’s the average age of a vehicle involved in a fatal crash in America.

The average age of vehicles on U.S. roads is 12.6 years, per S&P Global Mobility.^[2] The killing fleet is the fleet. Not ancient. Not new. Just… the cars people drive.

Break it down by era and the distribution sharpens into something uglier. Model years 2000 through 2009 account for 50.4% of all FARS fatalities—94,290 deaths out of 187,058—despite representing just ten vintages out of four decades of data.^[1] These are the vehicles built before the 2012 ESC mandate,^[3] before side curtain airbags became universal, before crash structures caught up with the physics of 4,500-pound crossovers hitting 3,000-pound sedans at 45 mph. Half the death toll comes from a single decade of cars that missed every major safety intervention of the modern era.

Pre-2000 vehicles—the actual old cars, the ones you’d expect to dominate the body count—contributed just 15% of fatalities. There simply aren’t enough 1998 Buick Centuries still on the road to move the needle, even with death rates that would horrify a modern safety engineer.^[1] Meanwhile, vehicles from 2020 and newer accounted for 3.2%. They haven’t been on the road long enough to accumulate a body count. The death machine has a sweet spot, and it’s parked in your neighbor’s driveway with a “2006” on the registration.

Individual models tell the story at higher resolution. The Chevrolet Cobalt concentrated 89% of its 1,540 FARS fatalities into just five model years (2005–2009)—its entire production run.^[1] The Cavalier: 69% in its worst five vintages. These were short-production vehicles with known defects (the Cobalt’s ignition switch killed at least 124 people per the DOJ settlement^[4]) that burned through their death inventory inside a single design generation. Contrast the Silverado: 32% in its worst five model years, spread across a production span reaching back to the 1980s. Long-production trucks dilute their fatality concentration across decades. Short-lived compacts concentrate it like a dose.

The policy implication grinds slowly. Congress mandated ESC for all new vehicles sold after September 2011.^[3] IIHS estimated ESC reduces fatal single-vehicle crash risk by 56% for SUVs and 49% for cars.^[5] Transformative. Except the median vehicle in a fatal crash is 12.5 years old, which means ESC didn’t reach the typical killing vehicle until approximately 2024. Thirteen years from mandate to median fleet penetration. Every safety regulation you celebrate today won’t reach the average fatal crash for over a decade.

Methodology

FARS_MODEL_YEAR data (323 vehicles, 2014–2023 crash years) provides fatality counts by vehicle model year. Average age at fatal crash computed as: sum of (2018.5 − model_year) × deaths, divided by total deaths, where 2018.5 is the midpoint of the 2014–2023 data window. This is an approximation—a 2005 model year vehicle could have crashed in 2014 (age 9) or 2023 (age 18); we use the midpoint. “Death concentration” = fraction of a model’s total fatalities occurring in its five highest-fatality model years. Era bins are non-overlapping: pre-2000, 2000–2004, 2005–2009, 2010–2014, 2015–2019, 2020+.

Limitations

FARS captures only fatal crashes—roughly 36,000 of the ~6.7 million annual crashes. The model-year distribution of fatal vehicles likely mirrors the model-year distribution of the on-road fleet; we cannot isolate an age-specific rate effect without per-vintage registration data, which FARS does not contain. The 12.5-year average may simply reflect the age profile of all cars on the road, not a disproportionate danger from middle-aged vehicles. Estimated VMT denominators introduce ±15% uncertainty for low-volume models. The “death concentration” metric is sensitive to production run length—a vehicle built for 5 years will mechanically concentrate deaths in fewer vintages than one built for 30.

Strongest Counterargument

The 12.5-year finding may be trivially explained by fleet composition. If the average car on the road is 12.6 years old, then the average car in a fatal crash being 12.5 years old tells us almost nothing about age-specific risk—it tells us what we already knew: most crashes involve the most common cars. The genuinely interesting question is whether 12-year-old vehicles crash at higher rates per mile than 5-year-old vehicles, and this analysis cannot answer that. NHTSA’s own research suggests vehicle age does increase crash risk due to degraded brakes, tires, and missing modern ADAS, but quantifying that effect requires registration-year-adjusted exposure denominators that don’t exist in FARS.^[1] The fleet age profile of death is real. Whether it’s meaningful beyond “most cars are middle-aged” remains an open question worth more data.