Traffic Deaths Fell 11% in Two Years. Truck Crash Injuries Rose 12%. That’s Not a Contradiction.

Fewer dead and more hurt, both true simultaneously, both explained by the same mechanism. It is hiding in the fleet composition data, and it explains why the death decline may be less triumphant than it looks.

Crash survivability is a vehicle-class function

I cross-tabulated FARS fatal crash data from 2014 through 2023 (337 models, 191,193 deaths, 336,202 crashes) and computed a ratio that NHTSA does not publish: crashes per death, broken down by vehicle class. A higher ratio means more crashes per fatality, which means higher occupant survivability.^[2]

Sports cars produce 1.47 crashes per death (68% lethal), sedans 1.55 (65%), SUVs 1.91 (52%), vans 1.95 (51%), and pickups 2.05 (49%).

A pickup occupant involved in a fatal crash event is 33% more likely to survive it than a sedan occupant, and an SUV occupant 25% more likely. These are not marginal differences; they are the spread between walking out of the ER and not walking out at all.

Now apply the fleet shift

Sedans represent 32.5% of the registered fleet but contribute 46.6% of all traffic deaths, a 1.44x overrepresentation. SUVs are 39.6% of the fleet but only 24.3% of deaths: 0.61x underrepresented. For every sedan replaced by a crossover on an American driveway, the national fatality count ticks down without a single crash being prevented.^[3]

Autoblog reported in 2025 that nearly four SUVs sold for every sedan, and that ratio keeps widening. As the fleet shifts from high-lethality sedans to high-survivability SUVs and pickups, deaths decline mechanically. The occupants who would have died in a Civic are surviving in a CR-V, but they are not surviving unscathed.

The injury signal nobody is tracking

NHTSA's own 2024 crash overview confirms the divergence: total fatalities down 4.3% (41,025 to 39,254) while truck-involved crash injuries climbed in both single-vehicle (+12%) and multi-vehicle (+8.2%) categories.^[1] Truck VMT did increase, which partially explains the injury rise. But the direction of both curves is exactly what a survivability-shift model predicts: more people walking away from crashes, fewer people walking away unhurt.

The strongest case against this thesis

Fatalities declined across all vehicle types in 2024, not just sedans. If fleet composition were the sole driver, sedan deaths should hold steady while SUV/pickup deaths fall. They didn't. AEB adoption, road design improvements, and post-pandemic speed normalization all contributed to the death decline, and this analysis cannot isolate how much is "survival shift" versus genuine crash reduction. IIHS research on vehicle size and weight confirms that heavier vehicles protect their own occupants better but transfer risk to lighter vehicles and pedestrians, complicating any simple narrative about "progress."^[4]

Limitations

FARS captures fatal crashes only, and our survivability ratio uses FARS crashes, which by definition involve at least one death, so we are measuring "conditional survivability given a fatal event," not total crash survivability. Model-level injury data does not exist in FARS, VMT estimates introduce roughly ±15% uncertainty, and we cannot separate whether SUV drivers are safer people or whether SUVs are safer machines. Both are probably true, and neither excuses the injury trajectory.

What you should do

If you drive a sedan built before 2018, you are in the highest-lethality segment of the fleet, which does not mean you should panic but does mean you should check your VIN at nhtsa.gov/recalls, confirm your vehicle has ESC (mandatory since 2012 model year, voluntary in many 2009+ models), and understand that the national fatality rate of 1.10 per 100 million VMT is not your rate. Your rate depends on what you drive, where you drive it, and whether the car next to you outweighs yours by 1,500 pounds.