America’s Death Rate Hit a Decade Low. Fifty-Nine Percent of Its Cars Went the Other Way.

NHTSA handed out its best quarterly number in twelve years last month. 0.99 deaths per 100 million vehicle miles traveled in Q1 2026, the lowest first-quarter fatality rate since 2014, and the kind of figure that ends up in press releases with words like “historic” and “turning point.”^[1] Full-year 2025 came in at the lowest death count since 2019. If 2026 holds pace, it will be the fifth consecutive year of declining road deaths in the United States. Real progress, measured in real bodies not on real roads.

So I did what seemed obvious: checked whether the individual cars on American roads are participating in this improvement. Pulled FARS data for every vehicle with enough model-year deaths to measure a trend. Compared exposure-adjusted fatality counts for model years 2010 through 2014 against model years 2017 through 2021. Controlled for the fact that a 2012 model year has been on the road for eleven FARS years while a 2021 has only three. Twenty-seven vehicles cleared the statistical threshold. Sixteen got worse.^[2]

Nissan Sentra leads the wrong direction, its 2017–2021 model years producing an exposure-adjusted death rate 90% higher than its 2010–2014 models, while the Jeep Wrangler climbed 85% and the Chevrolet Cruze rose 48% across the same generational comparison. Toyota Corolla, the bestselling nameplate in automotive history, shows a 43% increase in exposure-adjusted fatalities across model-year generations that were supposed to be safer.

Concentrated is the operative word: Dodge Ram dropped 60%, Chevrolet Impala dropped 59%, and Hyundai Sonata fell 37%, three vehicles that all received major redesigns with modern crash structures and standard stability control suites, pulling the average downward hard enough to mask what the majority of the fleet is doing. This is the same statistical mechanism that lets a stock index rise while most of its components fall, a handful of heavyweight winners dragging the headline number in a direction that misrepresents the broader portfolio.

Camry increased 17%, Honda Civic climbed 6%, Honda Accord rose 7%, and Ford’s F-150, America’s bestselling vehicle for four decades running, went up 4%. These are not exotic failures or statistical noise but rather the vehicles that comprise the operating majority of the U.S. passenger fleet, and their newer model years are killing at rates their predecessors did not.

What the national rate measures is the AVERAGE of all these trends weighted by fleet size and mileage, mixed with fleet turnover as older, deadlier vehicles age off the road and take their catastrophic death rates with them. A 2004 Ford Crown Victoria scrapped in 2022 removes a vehicle with a death rate several times the national average, and that subtraction improves the national number without a single newer car getting safer. Fleet retirement is accounting magic, effective and necessary and real, but it is not the same thing as the cars you can actually buy becoming less lethal.

Full-strength counterargument first: fleet expansion. A 2019 Nissan Sentra has more registered vehicles on the road than a 2012 Sentra because sales volumes shifted and fewer 2012s have been scrapped. More vehicles means more exposure, which means more deaths even at a constant rate per mile. Our exposure adjustment divides by FARS years, not by registered vehicle count per model year, which FARS does not provide. It is entirely possible that some vehicles in the “worse” column are actually maintaining the same per-vehicle risk and simply showing higher absolute numbers because more of them exist. We cannot rule that out with this data. Nobody can.

But the Sentra at +90% strains that explanation. Nissan sold approximately 200,000 Sentras per year through both eras, making a doubling from fleet expansion alone implausible. Wrangler at +85% occurred during a period when Jeep sales did grow substantially, which partially accounts for the increase, though not entirely. Corolla at +43% happened while Toyota’s annual Corolla volume was relatively stable. Fleet expansion is a confounder, not a complete explanation, and the question of what structural variable changed in how these vehicles interact with American driving behavior, road design, or crash dynamics remains stubbornly unanswered because nobody has tried to answer it with vehicle-level longitudinal data rather than national aggregates.

What this means for you: If you are shopping for a car and relying on the national fatality rate trend as evidence that newer equals safer, you are reading the wrong number. Check the IIHS crashworthiness ratings for your specific model and year. Look up the FARS fatality data for your vehicle at the Crash Report. A 2020 Nissan Sentra and a 2020 Hyundai Sonata occupy the same price bracket. Their model-year fatality trajectories point in opposite directions. A national average tells you nothing about either one.

Source: NHTSA FARS 2014–2023, NHTSA Q1 2026 preliminary estimates. Exposure adjustment divides model-year deaths by years of FARS exposure, not by registered fleet size, which FARS does not provide. Fleet expansion is a significant confounder for vehicles with growing sales volumes. Specific model-year rate claims reflect aggregate FARS fatal crashes and may not correspond to per-vehicle risk. See methodology for caveats.