24 Vehicle Models Kill Half of America. The Other 313 Split the Rest.

I sorted 337 vehicle models in the FARS fatal crash database by body count. Then I walked down the list, adding up deaths, waiting for the cumulative total to hit 50% of the 191,193 Americans who died in or around these vehicles between 2014 and 2023.^[1]

It took 24 models. Seven percent of the list.

The Silverado alone is 5% of the national body count: 9,591 dead. Ford's F-150 adds another 4.8%. The Honda Accord, a car your dentist drives, contributed 7,102 fatalities. By the time you reach the Impala at position 10, you have crossed 60,478 deaths on just ten nameplates.^[1]

Extend to 79 models (23.4% of the database) and you capture 80% of all deaths. The remaining 258 models, from the Mazda5 to the Ram 2500, split the last 20% among themselves. Traffic death is not distributed evenly. It is concentrated in a way that should terrify anyone who thinks "fix the worst offenders" is a viable national strategy.

Because there are two completely different lists of worst offenders, and they share almost nothing.

List A: raw body count. Silverado, F-150, Accord, Civic, Camry, Corolla, Altima, RAM, Explorer, Impala. These ten vehicles killed 60,478 people.^[1] They are also among the most common vehicles on American roads. Chevrolet put 5.7 million Silverados on the road. Ford: 6.6 million. Their death rates per 100 million VMT are 1.25 and 1.04, respectively. Moderate. Unremarkable. Toyota's Camry sits at 2.03. Your dentist's Accord clocks 3.07, which is elevated but hardly catastrophic for a sedan with a 45-year sales history that saturated the used-car market before electronic stability control existed.^[2]

List B: death rate. Hyundai Veloster (8.54 per 100M VMT), Chevy Tracker (7.83), Toyota Land Cruiser (6.27), Ford Mustang (6.02), Nissan Maxima (5.11), Chevy Cobalt (5.10), Chevy Impala (5.00), Chevy S-10 (4.83), Toyota Solara (4.25), Cadillac Seville (3.89).^[1] These are the vehicles that convert miles driven into funerals at the highest clip. Some are tiny (Veloster). Some are ancient (S-10, Seville). Some had known defects (Cobalt's ignition switch). All of them kill at rates 7 to 16 times the model-level median of 0.54.

Overlap between the two lists: one vehicle. The Impala. It manages the rare achievement of killing both the most people per mile and a staggering number in aggregate, thanks to its 656,000-unit fleet and a 5.0 rate that puts it in the company of sports cars and known deathtraps.

I computed the excess deaths from the top 10 rate killers. If each had performed at the model-level median rate of 0.54 per 100 million VMT, the total excess would be approximately 12,450 deaths eliminated over the decade.^[1][3] That is 6.5% of the total. Close to the 7.1% Pareto cutoff, but not identical. Concentration begets concentration, but the math does not hand you a clean policy lever.

Now consider the alternative intervention. A 10% reduction in the death rate of just the Silverado and F-150 would save approximately 1,879 lives over a decade. Eliminating the Veloster entirely saves 598. Retiring the Tracker, the Seville, and the Solara combined saves 1,889. The boring, incremental improvement of already-moderate vehicles outperforms the dramatic removal of dangerous ones.

This is the fundamental arithmetic trap of vehicle safety policy. Politicians want to ban the dangerous car. Advocates want to name and shame the deadliest model. But the models that kill the most people are not especially dangerous, and the models that are most dangerous do not kill many people. Pareto's law is real, but it offers no silver bullet. It offers two inadequate bullets fired in opposite directions.

For the math: I used FARS fatal crash records for 2014 through 2023, cross-referenced against estimated fleet sizes derived from industry sales data and assumed 15-year average vehicle lifespans, with VMT calculated at class-specific averages from the National Household Travel Survey.^[4] Rate equals deaths per 100 million vehicle miles traveled. Excess deaths equal actual deaths minus (VMT times median rate divided by 100).

The limitation that weakens everything above: Fleet estimates carry ±15% uncertainty for any individual model. VMT assumptions use class averages, not odometer readings. A Mustang driven 8,000 weekend miles and a Camry driven 18,000 commuter miles get the same per-mile denominator within their respective classes. This flattens rate differences for vehicles driven fewer miles and inflates them for high-mileage models. FARS also captures only fatal crashes. A vehicle with a low death rate might still be involved in thousands of injury crashes that never enter this dataset.

The counterargument: This entire Pareto distribution is trivially expected. Popular vehicles have more of everything: more sales, more miles, more crashes, more deaths. Of course the top 7% by volume captures 50% of deaths. You would find a similar distribution for repair costs, insurance claims, or parking tickets. The concentration tells you about market share, not about safety.

Correct. And that is precisely the point. Public debate treats "most deaths" as a safety metric. It is not. It is a popularity metric. The Silverado is not America's deadliest vehicle. It is America's most common vehicle that also kills people. Confusing the two has misdirected safety advocacy for decades.