919,000 Crashes. 2,872 Dead. Nobody Stayed.

Somebody hit a pedestrian in America last year. Then somebody drove away. This happened roughly 2,500 times per day.^[1]

A new AAA Foundation for Traffic Safety study, published March 11, 2026, put the full scope of hit-and-run in the United States into a single dataset for the first time in years. The numbers: 919,000 police-reported hit-and-run crashes in 2023. 242,000 injuries. 2,872 deaths. Hit-and-run now accounts for 15% of all police-reported crashes, the highest share ever recorded.^[1]

Fifteen percent. One in seven. You rear-end somebody in a parking lot, and the statistical odds that one of the other recent crashes in your ZIP code was a hit-and-run are now higher than at any point in American traffic history.

One in Four Pedestrians Left on the Road

Hit-and-run kills pedestrians at a wildly disproportionate rate. Over 70% of hit-and-run fatalities are pedestrians and cyclists, according to the AAA study. In 2023, one in four pedestrians killed in traffic was struck by a driver who fled.^[2]

That ratio has been climbing for a decade. IIHS Fatality Facts tracks it year by year:

• 2014: 953 pedestrian hit-and-run deaths (19% of pedestrian fatalities)
• 2017: 1,226 (20%)
• 2019: 1,310 (21%)
• 2020: 1,603 (24%)
• 2022: 1,939 (26%)
• 2023: 1,818 (25%)^[2]

Notice the jump. From 2019 to 2020, hit-and-run pedestrian deaths rose by 293 in a single year. Pre-COVID, the share had been inching up by a percentage point every two or three years. COVID compressed three years of drift into twelve months. It never drifted back.

The Profile of a Fleeing Driver

Less than half of hit-and-run drivers are ever identified.^[1] Among those who are caught, the AAA data paints a consistent portrait:

40% did not have a valid driver’s license. That connects directly to what we found in our unlicensed motorcycle investigation: people operating vehicles without authorization are dramatically overrepresented in fatal crashes, and they have a strong incentive to flee. If you’re already driving illegally, staying at the scene means criminal exposure beyond the crash itself.

Over 50% were driving vehicles not registered in their name. Borrowed cars, stolen vehicles, or cars registered to family members. Harder to trace. Easier to disappear.

Most crashed near their own ZIP code. These aren’t long-distance incidents. Most are local drivers, on local roads, in neighborhoods they know.

80% Happen in Darkness

Approximately 80% of hit-and-run fatalities occur in the dark.^{[1] [3]} NHTSA’s pedestrian safety data confirms the broader pattern: 77% of all pedestrian deaths happen in dark conditions, and 89% involve a single vehicle.^[3]

A pedestrian crossing a poorly lit arterial road at 10 PM. A single vehicle. No witnesses, or at least no witnesses with a plate number. Driver panics or calculates. Leaves.

We’ve covered the darkness infrastructure gap before: American road design prioritizes vehicle throughput over pedestrian visibility. Most fatal pedestrian corridors are arterials with 40-55 mph speed limits, infrequent crosswalks, and lighting designed for drivers, not for illuminating the people drivers might hit. The hit-and-run epidemic doesn’t exist separately from the darkness epidemic. They overlap almost perfectly.

Six Times the Deaths in the Poorest ZIP Codes

The AAA study broke hit-and-run fatalities by neighborhood vulnerability, using the CDC’s Social Vulnerability Index. The most vulnerable 20% of ZIP codes experience six times the hit-and-run fatalities of the least vulnerable 20%.^[1]

Six times. Not twice. Not three times. Six.

Middle-income ZIP codes fall at roughly three times the rate of the wealthiest. The gradient is steep and consistent. If you could pick one variable to predict where hit-and-run deaths cluster, household income would be more predictive than vehicle speed, road type, or time of day.

This tracks with our earlier investigation into safety as a luxury good. Wealthier neighborhoods get better lighting, lower speed limits, more crosswalks, more traffic-calming infrastructure, better police response times, and more surveillance cameras that help identify fleeing drivers. Poorer neighborhoods get arterial roads designed to move suburban commuters through at 45 mph. The pedestrians who live along those roads absorb the risk.

Only Three States Have a System for This

When a child is abducted, 50 states activate AMBER Alerts. When an elderly person with dementia wanders off, 37 states have Silver Alerts. When someone is killed in a hit-and-run, three states have a notification system: California, Colorado, and Maryland.^[1]

Three. They call them Yellow Alerts, and they function like AMBER Alerts for hit-and-run crashes: broadcast vehicle descriptions to highway message boards, phones, and media. The idea is simple enough. A fleeing driver has a vehicle. Somebody on the road might see it. Time matters.

2,872 people died in hit-and-runs last year. Forty-seven states have no broadcast notification mechanism to help catch the driver. Less than half of fleeing drivers are ever identified. It’s not that the technology doesn’t exist. It exists in Colorado.

COVID Broke Something. It Stayed Broken.

The 2020 spike is the critical inflection in this data. Hit-and-run deaths peaked at an all-time record of 2,972 in 2022. The share of crashes that were hit-and-runs jumped from roughly 12% pre-2020 to 15% post-2020.^[1]

Possible explanations for the 2020 spike: emptier roads encouraged faster driving. Pandemic stress increased impaired driving. DMV closures meant more unlicensed drivers. Police pulled back on traffic enforcement in many jurisdictions.

But explanations for why roads were empty don’t explain why the rate stayed elevated through 2021, 2022, and 2023, when traffic volumes returned to pre-pandemic levels. The strongest counterargument: increased dashcam and surveillance camera adoption means more crashes are now identified as hit-and-runs that would previously have been coded as “driver unknown.” Better detection, not worse behavior. But the AAA study uses consistent FARS coding methodology across the decade, and the timing of the spike — coinciding with COVID behavioral disruption rather than a gradual technology adoption curve — argues against a pure reporting artifact. Something behavioral shifted. The social contract around staying at the scene of a crash weakened during COVID, and it has not recovered.

This is the pattern we keep finding in traffic safety data. COVID didn’t cause new problems so much as accelerate existing ones past a threshold they couldn’t retreat from. Unlicensed motorcycle deaths jumped during COVID and stayed up. Pedestrian fatalities jumped during COVID and stayed up. Now hit-and-run rates jumped during COVID and stayed up.

The Math Nobody Does

Here is a calculation the AAA study does not perform, so we will.

If hit-and-run pedestrian deaths had held at the 2019 rate of 21% instead of rising to 25%, approximately 293 fewer pedestrians would have died in hit-and-runs in 2023. Those people were not killed by a new road design or a new vehicle technology. They were killed by a behavioral shift: more drivers deciding to leave.

Put differently: the 2020 behavioral change, sustained over four years, has added roughly 1,100 excess hit-and-run pedestrian deaths to the American total. That’s not an engineering problem. That’s not a vehicle design problem. That’s a people-decided-the-consequences-aren’t-real problem.

And given that fewer than half of fleeing drivers are identified, the consequences largely aren’t real. The expected penalty for killing a pedestrian and driving away is, in most American jurisdictions, approximately nothing.

Methodology note: Primary data from AAA Foundation for Traffic Safety, “Understanding the Increase in Fatal Hit-and-Run Crashes: Prevalences of Crashes, Injuries, and Deaths in the United States, 2017–2023,” published March 11, 2026. Pedestrian hit-and-run year-by-year data from IIHS Fatality Facts: Pedestrians, 2023. Additional pedestrian fatality context from NHTSA Traffic Safety Marketing. The “excess deaths” calculation applies the 2019 hit-and-run share (21%) to the 2023 total pedestrian fatality count (7,314) to estimate a counterfactual of 1,536 hit-and-run pedestrian deaths vs. the observed 1,818, yielding a difference of 282. We use “approximately 293” elsewhere in the text, which reflects the raw year-over-year change from 2019 to 2020 (1,310 to 1,603); the counterfactual calculation yields a slightly different figure because it applies the 2019 share to the 2023 total rather than tracking absolute year-over-year change. The 6x ZIP code disparity uses CDC Social Vulnerability Index quintiles as reported by AAA; individual ZIP code variation may be larger. Hit-and-run identification rates vary significantly by jurisdiction and are not uniformly reported across states; the “less than half” figure is AAA’s national estimate. FARS captures only fatal crashes; hit-and-run crashes resulting in injury or property damage only are tracked by state-level crash databases with inconsistent reporting standards.