The U.S. Death Rate Hit 1.10. For 23 Million Vehicles, It’s Still 2.75.

NHTSA announced it in January with the quiet satisfaction of an agency that needed a win: 36,640 traffic deaths in 2025, a rate of 1.10 per 100 million vehicle miles traveled, the lowest since 2014.^[1] Fifteen consecutive quarters of decline. The National Safety Council's February 2026 preliminary data pushed even further, landing at 1.01 per 100M VMT.^[2] Sub-1.0 felt imminent. Progress. Finally.

Then we decomposed the FARS database by per-model fatality rate and fleet share, and the celebration stopped making sense.

Sort all 337 FARS-tracked models into three buckets by their individual fatality rates, and America's vehicle fleet fractures into populations so distinct they might as well occupy different countries. The safe fleet (241 models, rate below 1.0 per 100M VMT) covers 109 million registered vehicles and runs at an effective rate of 0.40. These vehicles passed the 1.10 milestone years ago. RAV4 at 0.19, Ram 1500 at 0.13, Equinox at 0.36.^[3] For these drivers, NHTSA's announcement was confirmation of something their crumple zones already knew.

The danger fleet sits at the other end: 37 models, 23 million registered vehicles, an effective fatality rate of 2.75 per 100M VMT. That rate is 2.5 times the national figure everyone celebrated in January. These vehicles hold 13.4% of the fleet but generate 38.2% of all FARS deaths: 73,048 fatalities across the 2014-2023 reporting window.^[3] For a driver in a danger-fleet vehicle, the national rate dropping to 1.10 changed absolutely nothing about the first 150 milliseconds of their next crash.

And the composition of the danger fleet is the part nobody expected. Twenty-one of those 37 models are sedans, and their names read like a used-car lot's bestseller list: Honda Accord at 3.07 per 100M VMT and 7,102 deaths, Civic at 2.25 and 6,553, Toyota Camry at 2.03 and 6,328, Nissan Altima at 2.88 and 4,787, Chevy Impala at 5.0 and 3,774.^[3] The vehicles killing people at nearly triple the national rate are not the oversized body-on-frame trucks that safety advocates have spent a decade campaigning against. They are the most ordinary sedans in America, the cars sitting in every apartment complex parking lot and community college overflow lot in the country, driven by the people who can least afford to replace them.

A 59-model middle fleet (rate 1.0 to 2.0, 39.5 million vehicles) bridges the gap. Its effective rate of 1.30 sits close enough to the national average to show where the 1.10 figure actually comes from: it is the weighted mean of a safe fleet that was already at 0.40 and a danger fleet that hasn't budged from 2.75, with enough middle-fleet mass to pull the average down to something reportable.^[3]

Lethality compounds the divide. In the safe fleet, 51% of fatal crashes kill the vehicle's driver. In the danger fleet, that conversion rate reaches 65%.^[3] Same crash forces, different survival odds, because the structural engineering separating a 2008 Altima from a 2023 Rogue is not incremental. It is a generation of side-impact protection, high-strength steel pillars, curtain airbags that actually cover the rear, and electronic stability control calibrations refined through millions of real-world miles of data that the older platform never received and never will.

The strongest objection writes itself: those danger-fleet sedans are overwhelmingly older model years whose replacements already sit in the safe fleet. The 2024 Civic is a fundamentally different structure than the 2009 Civic carrying the death-rate burden. Natural fleet turnover will solve this, or so the argument goes. Maybe. S&P Global Mobility pegs the average U.S. vehicle age at 12.6 years and climbing.^[4] The danger fleet is not cycling out. It is aging in place, accumulating miles in the driveways of owners who bought used precisely because new was unaffordable, and the safety gap between what they can afford and what NHTSA's rate describes widens every year the national number drops.

We cannot tell you whether the three-fleet gap is widening or narrowing over time; FARS provides cumulative 10-year rates, not annual snapshots by model. Fleet size estimates rely on sales data and scrappage assumptions, not DMV registration counts. VMT per vehicle is estimated within plus or minus 15% for low-volume models.^[3] The uncertainty doesn't erase the pattern. It means the 6.9x gap could be 5.5x or 8.3x, and neither version changes the conclusion.

What to do: If you drive a Honda Accord (any generation before 2023), Civic (pre-2022), Nissan Altima, or Chevy Impala, your personal fatality risk is 2.5 to 4.5 times the national average that everyone keeps quoting. Check your VIN at nhtsa.gov/recalls for open campaigns, but understand that no recall addresses the structural deficit. If replacing the vehicle is not an option, the two highest-value interventions are tires (the only contact patch between you and 2.75) and seatbelt use (FARS consistently shows unbelted occupants in older sedans at rates double the fleet average). The cheapest way to exit the danger fleet is a used RAV4 or CR-V from 2018 or newer. At 0.19 and 0.31 respectively, they occupy a different country entirely.

Source: NHTSA FARS 2014–2023. Fatality rates are 10-year cumulative estimates, not single-year snapshots. Fleet sizes derived from sales data and scrappage models, not registration counts. VMT per vehicle estimated (±15% for low-volume models). See methodology for caveats.