The Fleet Retirement Dividend: America's Deadliest Vehicle Generation Is Finally Dying

I ran a cross-tabulation on FARS model-year cohorts last week, and the number that came back was so lopsided I reran it twice. Vehicles built between 2000 and 2006 are responsible for 69,625 fatalities in the FARS database across 2014–2023.^[1] Not injuries. Deaths. One seven-year production window, more fatal than any other era in the dataset by a factor of two.

Now those vehicles are finally leaving the road, and the national fatality numbers are responding. NHTSA's early estimates for the first half of 2025 show 17,140 deaths, down 8.2% from 18,680 in the same period of 2024.^[2] The fatality rate per 100 million vehicle miles traveled dropped to 1.06, the lowest mid-year figure since 2014. Thirteen consecutive quarterly declines. Press releases credit enforcement campaigns and safety technology. Nobody mentions fleet composition.

They should. In absolute terms, the math is straightforward. Model year 2005 alone produced 11,363 FARS deaths across all vehicle types. For context, the entire 2017-and-newer cohort managed 19,726. A single pre-ESC model year, outpacing six years of modern vehicles equipped with electronic stability control, automatic emergency braking, and blind-spot monitoring. The 2001 Ford F-150 racked up 672 deaths from just that one production year. The 2004 Chevrolet Silverado hit 663.^[1]

Why This Cohort Kills

Three factors converge. First, volume: 2000–2006 represented peak US auto sales, roughly 17 million units annually, creating the largest single-era fleet in American history. Second, technology: none of these vehicles were required to carry electronic stability control, which IIHS later found reduces single-vehicle fatal crashes by 56% and fatal rollovers by 70–90%.^[3] The ESC mandate didn't phase in until 2007, with full compliance required by September 2011. Third, economics: by the time our FARS window opens in 2014, these vehicles are 8–14 years old, cheap enough to become the default transportation for younger and lower-income drivers who accumulate the most exposure-adjusted risk.

Compare the eras in our FARS data:

That 2000–2006 band is wider than everything else combined from 2007 onward. It is, by a comfortable margin, the deadliest vehicle generation ever measured by FARS.

The Scrappage Clock

S&P Global Mobility reports the average US vehicle age hit a record 12.8 years in 2025, with passenger cars averaging 14.5 years.^[4] At a 4.5% annual scrappage rate, a vehicle cohort roughly halves every 15 years. A model year 2000 vehicle has a survival probability around 32% in 2025. Model year 2006 sits closer to 42%. Five years ago, when fatalities were still climbing post-COVID, those same figures were 55–65%.

Run the replacement math: if the 2000–2006 cohort accounted for roughly 25% of VMT in 2020 and closer to 12% in 2025, and their per-VMT fatality rate was approximately double the post-2012 fleet due to absent ESC and older crash structures, then fleet composition alone could explain 3–5 percentage points of the national fatality decline. That's not everything. But it's a substantial chunk of NHTSA's celebrated 8.2% drop, and nobody in the press conference mentioned it.

Limitations

This is a floor estimate, not a ceiling. Our FARS data covers 2014–2023; we don't have 2024–2025 breakdowns by model year yet. Scrappage estimates use aggregate rates, not vehicle-specific survival curves. Pickups, which constitute a disproportionate share of the deadliest cohort, tend to survive longer than sedans, so the 2000–2006 pickup fleet may be retiring more slowly than the overall rate suggests. We also cannot isolate fleet composition from behavioral changes: seatbelt use continues to climb, rideshare has reduced some drunk driving, and EMS response times have improved in urban areas.

The Counterargument at Full Strength

The strongest case against the fleet-composition thesis: the shift to larger vehicles. New cars in 2025 are overwhelmingly SUVs and crossovers. They are safer for their occupants, but IIHS research on vehicle size and weight shows they transfer crash energy to smaller vehicles and pedestrians.^[5] Pedestrian deaths are at their highest levels since the 1980s. It is entirely possible that the "fleet retirement dividend" for occupant deaths is being partially offset by an "aggressor vehicle tax" on vulnerable road users. The net benefit to the overall fatality number could be smaller than the compositional math implies.

Still. Sixty-nine thousand, six hundred and twenty-five deaths from seven model years. That number should be tattooed on every vehicle design spec from that era. The good news is those vehicles are dying. The bad news is it took 20 years for the scrappage clock to do what regulators could have done in 2001 by mandating ESC a decade earlier.