The Unbuckled 8%: A Tiny Minority Produces Nearly Half of America’s Traffic Deaths
I ran the numbers twice because the ratio looked like a spreadsheet error.
NHTSA’s 2024 National Occupant Protection Use Survey counted front-seat seatbelt usage at 91.2%.[1] That means 8.8% of front-seat occupants ride without a belt. One in eleven. A small, stubborn fraction of the driving population that decides every single trip that the three-point harness isn’t worth the two seconds.
That 8.8% produced 10,484 of the 24,238 passenger vehicle occupant deaths in 2023. Forty-four percent of the body count from fewer than nine percent of the riders.[2] The year before was worse: 11,410 unrestrained deaths, 45% of the total.[3]
That’s not a risk factor. That’s a population-level death sentence with a simple opt-out that 91% of Americans already use. (A caveat: the 8.8% comes from NOPUS 2024 observational surveys at intersections; the 44% comes from FARS 2023 fatality data. They measure different populations in different years. The ratio is illustrative, not a precise attributable fraction.)
The pickup truck gap
Not all vehicles are equal in the unbuckled-death calculation. NOPUS breaks seatbelt use by vehicle type, and pickups sit at the bottom: 86.4% belt use, compared to 92.8% for passenger cars and 92.1% for SUVs.[1]
That 6-point gap between pickups and everything else compounds in FARS. Pickup occupants accounted for 27% of all unrestrained fatalities in 2023 despite representing roughly 18% of registered passenger vehicles (per IHS Markit registration data).[2][7] The math is straightforward: fewer belts plus higher fatality rates when unbelted (NHTSA estimates seatbelts reduce fatal injury risk by 60% in light trucks versus 45% in cars)[4] equals a disproportionate share of the morgue.
This tracks with what we found in the hood-height investigation last week. The same vehicle class that’s killing more pedestrians outside the truck is also killing more of its own occupants inside it, in part because a meaningful fraction of those occupants won’t buckle up.
After midnight, the numbers get ugly
Split FARS fatality records by time of day and restraint status and a pattern jumps off the screen. Unrestrained fatalities are roughly three times more concentrated between 9 PM and 6 AM than during daytime hours, based on the FARS HOUR field cross-tabulated with restraint use.[2] The ratio compares unrestrained deaths per hour in nighttime versus daytime windows.
The nighttime concentration isn’t random. It overlaps almost perfectly with alcohol-impaired driving. NHTSA reports that seatbelt use among alcohol-positive drivers in fatal crashes is approximately 10 percentage points lower than among sober drivers.[5] In 2022, drunk driving killed 13,524 people, 31.8% of all crash deaths. The Venn diagram of “drunk,” “unbuckled,” and “dead” has a massive overlap zone, and it peaks after midnight.
| Risk Factor | Share of Passenger Vehicle Fatalities |
|---|---|
| Unrestrained (any time) | 44% |
| Alcohol-impaired (BAC ≥0.08) | ~32% |
| Unrestrained + nighttime | ~29% |
| Unrestrained + alcohol-positive | ~18% |
The overlap matters because it complicates the causal story. Someone who dies unbuckled at 2 AM going 95 in a pickup didn’t die from one bad decision. They died from a cluster of them. The seatbelt was the last line of defense they chose to skip.
The ejection multiplier
IIHS provides the most visceral data point in this entire analysis. Unbelted occupants have a complete ejection rate of roughly 20% in fatal crashes. Belted occupants: 0.03%.[6]
That’s not a percentage-point difference. That’s a 667x multiplier. And ejection is nearly always fatal. IIHS estimates ejected occupants face roughly 91 times the fatality risk of those who stay in the vehicle.
Put it in practical terms: a belted driver who rolls a pickup at highway speed has a bad day and probably a hospital stay. An unbelted driver in the same crash exits the vehicle through the windshield or side window and hits the pavement at whatever speed the tumbling truck was going. The belt doesn’t prevent the crash. It prevents the occupant from becoming a projectile.
Who doesn’t buckle up
NHTSA published a demographic study in 2020 analyzing the characteristics of occasional and never-users.[5] The composite portrait:
- Male: three times more likely to skip the belt than female
- Age 16–24: lowest use rate of any age bracket
- Rural: 10% less likely to buckle than urban drivers
- Pickup truck: 86.4% use vs. 92%+ for cars and SUVs
- Rear seat: only 78% usage (but front seat is where most deaths occur)
There’s also a strong correlation with other risk behaviors. Non-users are more likely to speed, more likely to drive impaired, and less likely to use child restraints for passengers. The seatbelt isn’t just a safety device. It’s a sentinel indicator. Seeing an unbuckled driver tells you more about their risk profile than any single data point should.
The counterargument worth taking seriously
The honest objection to the 8%-produces-44% framing: selection bias. People who skip seatbelts are the same people who speed, drink, and drive aggressively. The belt isn’t the variable that killed them. It’s a proxy for a risk-tolerance personality cluster that would produce elevated fatalities regardless.
This is partially true. Some fraction of unrestrained fatalities would have died belted because the crash was unsurvivable at any restraint level. High-speed single-vehicle rollovers on rural highways at 2 AM are often fatal even with a belt.
But the controlled efficacy data survives this objection. NHTSA estimates, based on matched-pair studies controlling for crash severity, that seatbelts saved 14,955 lives in 2017 alone.[4] The 45% risk reduction for car occupants and 60% for light-truck occupants comes from studies that hold crash energy constant. Applying those efficacy rates to the 2023 unrestrained fatalities suggests seatbelts could have saved roughly 4,700 to 6,300 of the 10,484 who died without one. Not all, because some crashes are unsurvivable at any restraint level. But most. The belt works even when the driver is making every other wrong choice. That’s the point of a passive safety device. It doesn’t require good judgment. It only requires one click.
New Hampshire and the enforcement gap
Forty-nine states plus D.C. and five territories have mandatory seatbelt laws. New Hampshire exempts adults over 18. Its 2024 front-seat belt use rate: 75.5%, dead last in the nation.[1]
Among the 49 states that do mandate belts, 34 have primary enforcement (an officer can pull you over for the belt alone) and 15 have secondary enforcement (can only ticket it if they stop you for something else). Primary-law states consistently report 5–10 percentage points higher usage.[1]
The enforcement type matters more than the fine amount. A $25 primary-enforcement ticket moves behavior more than a $200 secondary-enforcement one because the probability of getting caught, not the penalty, drives compliance. This is basic deterrence theory, and the seatbelt data confirms it at a national scale.
14,955 saves. 10,484 refusals.
Seatbelts saved almost 15,000 lives in 2017. In 2023, about 10,500 people died without one on. Those two numbers sit next to each other in my spreadsheet and I keep looking back and forth between them.
The belt is the single most effective safety intervention in a car. More than airbags (which assume a belt to function correctly). More than ESC (which prevents crashes but can’t help mid-crash). More than AEB, crumple zones, or any of the engineering marvels we write about on this site. A three-point harness invented in 1958 by a Volvo engineer who gave the patent away for free still outperforms every subsequent innovation in raw lives saved per dollar.
And 8.8% of front-seat occupants look at it every time they sit down and decide: no.
Methodology note
FARS captures only crashes involving at least one fatality, so the 44% figure reflects unrestrained share of deaths, not of all crashes. Seatbelt usage rates come from NOPUS, which observes front-seat occupants at controlled intersection sites and may not perfectly represent highway or rural driving populations where non-use is higher. The 8.8% non-use figure is a national average; state-level rates range from 75.5% (New Hampshire) to 96%+ (California, Hawaii). The demographic composite is from a 2020 NHTSA report and may not fully reflect post-pandemic shifts. Alcohol-involvement overlap percentages are estimated from FARS cross-tabulations where both restraint and toxicology fields are populated; not all fatal crash records have complete data in both fields. The 14,955 lives-saved estimate uses NHTSA’s standard effectiveness methodology, which models expected fatalities without belts and subtracts observed belted fatalities.