How Many People Die Running Red Lights Each Year?

Quick answer: In 2024, 1,119 people were killed in crashes that involved red light running in the United States, according to the Insurance Institute for Highway Safety (IIHS). More than 141,000 people were injured in red light running crashes that year. Half of those killed were pedestrians, bicyclists and people in other vehicles who were hit by the red light runners.

The headline numbers

The IIHS topic page on red light running, compiled by senior research transportation engineer Wen Hu, gives the national counts for 2024:

• 1,119 people killed in crashes that involved red light running in 2024
• More than 141,000 people injured in red light running crashes in 2024
• Half of those killed were pedestrians, bicyclists and people in other vehicles who were hit by the red light runners

That last figure is the central fact about this kind of crash. In half of the deaths, the person killed was not the driver who ran the light. They were someone crossing the road, riding a bicycle, or travelling in another vehicle that was hit.

Red light running is defined narrowly: a driver commits a violation if a vehicle enters an intersection after the signal has turned red. A driver already in the intersection when the signal changes, for example while waiting to turn left, is not a red light runner.

It is also frequent. At five busy intersections in Fairfax, Virginia, before cameras were used, a motorist ran a red light every 20 minutes at each intersection on average. Across 19 intersections without cameras in four states, the rate was 3.2 violations per hour per intersection. Readers can compare state figures on the United States state pages, read other data articles in all articles, and see the latest reported crashes on accidents today.

Who runs red lights

Among drivers involved in fatal red-light-running multiple-vehicle crashes in 2024, the red light runners were more likely than other drivers to be male and younger, to have prior crashes or alcohol-impaired driving convictions, to have been speeding or alcohol-impaired at the time, and less likely to have held a valid driving licence.

The gap between what drivers say and what they do

A 2024 national telephone survey by the AAA Foundation for Traffic Safety, published in 2025 and reported by IIHS, found a clear gap between attitudes and behaviour:

• 80% of drivers said it is very or extremely dangerous to drive through a light that had just turned red when they could have stopped safely
• 27% reported doing exactly that in the past 30 days

In other words, four in five drivers recognise the danger, yet more than one in four admitted to the behaviour in the previous month. Drivers already know that running a fresh red light is dangerous. The evidence points to two things that stop them doing it: enough yellow time to stop safely, and a real expectation of being caught.

What reduces red-light running

Signal timing comes first. Increasing yellow timing in line with Institute of Transportation Engineers guidelines can significantly decrease violations and reduce total crashes, injury crashes and right-angle crashes, according to studies reported by IIHS.

A Philadelphia study tested timing and cameras in sequence. Lengthening the yellow signal by about a second reduced violations by 36%. Adding camera enforcement then cut violations by a further 96%. Cameras work best where signal timing is already fair.

An IIHS study comparing large cities with cameras with those without found that cameras reduced the fatal red light running crash rate by 21% and the rate of all types of fatal crashes at signalised intersections by 14%. The Campbell Collaboration reviewed 38 controlled before-and-after studies and estimated a 20% reduction in all injury crashes and a 29% reduction in right-angle injury crashes.

Removing cameras has the opposite effect. In 14 cities that shut down their programmes during 2010 to 2014, the fatal red light running crash rate was 30% higher than expected if the cameras had been left on, and the fatal crash rate at signalised intersections was 16% higher. In Houston, which turned off its cameras in 2011, deactivation was associated with a 23% increase in right-angle crashes at the intersections that had previously had cameras.

There is a trade-off. Some studies have reported that cameras can increase rear-end crashes, as drivers brake sharply at lights. A Federal Highway Administration study of camera programmes in seven cities found that right-angle crashes decreased by 25% while rear-end collisions increased by 15%. It found a positive aggregate economic benefit of more than 18.5 million dollars across the seven communities, because rear-end crashes tend to be much less severe than front-into-side crashes. Automatic emergency braking, now standard on nearly all new vehicles, should shrink the rear-end increase as that technology spreads.

Cameras in use, and public support

The first camera programme was implemented in 1992 in New York City. During 2025, 351 United States communities operated red light safety camera programmes, according to information tracked by IIHS. The total has declined since 2012, because more programmes have been discontinued than initiated.

Major cities with cameras include Chicago, Denver, New Orleans, New York City, Philadelphia, San Francisco, Seattle and Washington, D.C. A 2011 IIHS survey in 14 big cities with longstanding camera programmes found that two-thirds of drivers supported their use. A 2012 survey in Washington, D.C. found that 87% of residents supported them.

How these numbers were counted

All death and injury counts come from the IIHS topic page on red light running, which reports crashes that involved red light running in 2024, the latest year on that page. The attitude figures come from a 2024 national telephone survey by the AAA Foundation for Traffic Safety, published in 2025. Camera figures come from IIHS city comparisons and reviews of controlled before-and-after studies cited on the page. Figures from different studies cover different cities, periods and crash types, so they should not be added together.