Most people assume traffic lights go where traffic is bad. That's not wrong exactly, but it's not how it works. Cities follow a federal document called the Manual on Uniform Traffic Control Devices — the MUTCD — which sets out specific numerical thresholds a location must meet before a signal can be installed. The thresholds are called "warrants."

There are nine warrants in the MUTCD. An intersection only needs to satisfy one to qualify for a signal, but engineers typically study several before making a recommendation.

Quick answer

A traffic signal usually starts with a warrant study: vehicle counts, pedestrian crossings, crashes, delay, school crossings, or network needs. Meeting a warrant does not automatically require a signal. It means engineers can consider one, then compare it with alternatives such as stop control, roundabouts, pedestrian islands, or flashing beacons.

The warrant that matters most: vehicle volume

Warrant 1 is Eight-Hour Vehicular Volume. Engineers count the number of vehicles passing through the intersection every hour for a full day and pull out the eight worst hours. Those numbers are then compared against a chart that accounts for the number of lanes and the speed limit. If the actual counts exceed the chart thresholds in those eight hours, the warrant is met.

The thresholds aren't arbitrary. They're based on decades of research into when a signal actually reduces delay and conflict versus when it just adds unnecessary stops to a road that was flowing fine.

The pedestrian warrant

Warrant 4 covers pedestrian volume. If more than 100 pedestrians per hour cross a leg of the intersection during any four hours of the day — and vehicle volume makes crossing difficult — the pedestrian warrant is met. This is how school crossings near busy roads end up with signals even when the vehicle volumes alone wouldn't justify one.

What happens after a warrant is met

Meeting a warrant doesn't automatically mean a light goes in. It means a signal may be warranted. The engineer then considers whether the signal would actually improve safety and flow, or whether a roundabout, a stop sign, or a pedestrian refuge island would do the job better with less cost and fewer rear-end collisions (which signals notoriously increase compared to other intersection controls).

The unintended consequences nobody talks about

Signals create what traffic engineers call "platooning" — vehicles bunch up behind red lights and move in waves. This can actually increase congestion downstream because the road alternates between empty and suddenly full. Roundabouts eliminate this entirely because flow is continuous, which is why most traffic engineers prefer them for lower-volume intersections where warrants are marginal.

The other issue is the "Washington effect" — communities that get a signal tend to report more accidents afterward, not fewer. This isn't because the signal made things worse. It's because locations are usually studied after a serious accident, so there's a selection bias toward places that were already dangerous. The signal often brings numbers back to normal, but normal looks like an improvement relative to the spike that triggered the study.

The political dimension

Traffic engineers will tell you off the record that a significant percentage of signals in any city shouldn't be there. They were installed because a council member got constituent complaints, or a developer demanded access to a new subdivision. The MUTCD warrants are a technical standard, but engineers are employees of cities run by elected officials.

The result is a lot of intersections with signals that primarily serve to make drivers stop when they don't need to, which trains drivers to ignore the signals, which makes the ones that actually matter less effective.

Infrastructure at scale is always partly technical and partly political. Traffic lights are a good example of a case where the technical answer and the political answer diverge fairly often.

For more infrastructure systems, read how the power grid stays balanced, how water gets to your tap, and how airport baggage works.

Sources and further reading