Road safety

The US death rate from motor vehicle traffic crashes fell 31% from 14.9 per 100,000 in 2003 to 10.3 in 2014, rose 29% to 13.3 in 2021, and then fell to 12.5 in 2023, when 43,273 people died [1]. Over the same period, drug overdoses overtook traffic crashes as the leading cause of death from unintentional injury [1].

Road users differ greatly in risk. In 2024, 6,228 motorcyclists were killed, 16% of all US traffic deaths, and per vehicle mile travelled motorcyclists were almost 27 times as likely as passenger car occupants to die in a crash and almost 5 times as likely to be injured [9]. Among the 22,713 passenger vehicle occupants killed in 2024, 48% were not wearing a seat belt where use was known [3].

The Second Strategic Highway Research Program Naturalistic Driving Study fitted cameras and sensors to the vehicles of more than 3,500 drivers for three years, recording more than 35 million miles of driving. Researchers compared the seconds before 905 injury and property-damage crashes with randomly sampled ordinary driving. Driver-related factors (error, impairment, fatigue or distraction) were present in almost 90% of crashes, some form of error in 73.7% and observable distraction in 68.3% [2].

Crash risk was measured against attentive, sober, alert driving. Observable drug or alcohol impairment was associated with about 35.9 times the risk, and judgement errors such as speeding well above the limit or driving too fast for conditions with 11.1 times the risk. Distraction of any kind doubled the risk, and drowsiness or fatigue was associated with 3.4 times the risk [2]. The authors noted that the impairment cases were probably the more severe ones, visible on video, so the true prevalence of impairment is likely higher and its average risk somewhat lower [2].

Full deep dive: road safety.

Speed raises both the chance of a crash and the energy released in one, because kinetic energy rises with the square of speed. For more than two decades speeding has been involved in about a third of US traffic deaths, and in 2024 it was a contributing factor in 29%, or 11,288 deaths [4].

About 30% of US traffic deaths involve a drunk driver with a blood alcohol concentration of 0.08 g/dL or higher; 11,904 people died in such crashes in 2024 [5]. In 2024, 29% of drivers aged 15 to 20 who died in crashes had a blood alcohol concentration of 0.01 g/dL or higher [5]. The wider health effects of drinking are described under alcohol.

Cannabis raises crash risk less than alcohol. A 2016 meta-analysis of 28 estimates from 21 observational studies found that acute cannabis intoxication was associated with a statistically significant increase of low to moderate size (odds 1.36 times as high in one model and 1.22 times in another), and noted that remaining biases in the studies may limit a causal interpretation [10].

A case-crossover study (in which each driver's phone use just before the crash is compared with the same driver's use on earlier trips) of 456 drivers treated in hospital after crashes in Perth, Australia, checked their accounts against phone company records. Phone use in the 10 minutes before a crash was associated with about four times the likelihood of crashing (odds ratio 4.1, 95% CI 2.2 to 7.7). The risk was raised whether or not a hands-free device was used: 3.8 times with hands-free and 4.9 times with handheld phones [11].

The naturalistic driving study found that activities taking the eyes off the road carried the highest risks: dialling a handheld phone (12.2 times), reading or writing (9.9 times), reaching for an object other than a phone (9.1 times) and texting (6.1 times). Talking on a handheld phone was associated with 2.2 times the risk and any handheld phone use with 3.6 times [2].

A study of drivers in a representative sample of US crashes compared the sleep reported for the previous 24 hours by drivers whom crash investigators judged to have contributed to the crash with that of drivers who did not contribute, who were taken to represent all drivers on the road. Compared with 7 hours or more, 6 to 7 hours of sleep was associated with 1.3 times the crash rate, 5 to 6 hours with 1.9 times, 4 to 5 hours with 4.3 times and less than 4 hours with 11.5 times [6]. The author noted that the risk after 4 to 5 hours of sleep is similar to that estimated for driving with a blood alcohol concentration at or slightly over the US legal limit of 0.08 [6].

A related peer-reviewed analysis of 6,845 drivers found that those who had slept less than 4 hours had 15.1 times the odds (95% CI 4.2 to 54.4) of being responsible for their crash, compared with drivers who had slept 7 to 9 hours, with the excess concentrated in single-vehicle crashes [12]. Sleep was self-reported in both analyses. Sleep regulation is described under how sleep works.

Seat belts reduce the risk of fatal injury by 45% and of moderate to critical injury by 50% for front-seat passenger car occupants, and by 60% and 65% in light trucks [3]. US seat belt use was 91.3% in 2025, and seat belts saved an estimated 14,955 lives in 2017 [3]. Unbelted occupants remain overrepresented among deaths: 48% of passenger vehicle occupants killed in 2024 were unrestrained, rising to 56% among those killed at night [3].

A Cochrane review of 61 observational studies found motorcycle helmets reduced the risk of death in a crash by an estimated 42% (odds ratio 0.58, 95% CI 0.50 to 0.68) and of head injury by 69% [7]. A 2017 meta-analysis of 40 studies of more than 64,000 injured cyclists found that helmet use was associated with lower odds of head injury (0.49), serious head injury (0.31) and fatal head injury (0.35), with no clear association with neck injury [8]. Cycling as exercise is described under sports and active hobbies.

Front crash prevention systems use cameras, radar or lidar to detect an impending collision, warn the driver and, in many vehicles, brake automatically. Studies by the Insurance Institute for Highway Safety found that systems combining forward collision warning with automatic braking cut rear-end crashes in half, while warning alone reduced them by 27%, and that automatic braking systems that recognise pedestrians cut pedestrian crashes by 27% [13]. Automatic braking produced larger reductions in rear-end crashes with injuries than in rear-end crashes overall, suggesting that it also lowers impact speed in crashes it does not prevent [13].