What heat costs runners

What heat costs, by distance

Measured July 2026. Within-athlete effect per +5°C of race-morning heat stress.

Read each row as: for every 5°C hotter the race morning, finishers at that distance ran this much slower, measured on the same runners across their cooler and warmer races at that distance, so it's the heat talking, not a different crowd. The effect is invisible at 5K, barely there at 10K, and takes hold over the longer distances: the longer you are out on the course, the more the heat compounds.

These are ranges, not exact numbers. Each one shifts depending on whether we compare a runner only on the very same course, or across all their races at that distance. We show both ends rather than pick one. The marathon is the least certain, since so few runners repeat one.

Look up your own race

Pick a race you've run more than once and see how its editions compare. When enough people have run it across several years, the table follows those repeat runners and shows how much slower each edition was than their own usual time. That isolates the weather from who happened to show up.

When too few people have repeated it, the table falls back to the whole field's median finish. That can't separate the weather from the crowd, since a slower year might just mean a slower field turned up, so treat it as a rough guide. The 🔥 marks the hottest edition.

How the number is measured

We never compare a fast runner on a cool day to a slow runner on a hot day; that would just measure who showed up. Instead we pair each race with the same runner's previous race at that distance — whichever event that was, up to 18 months earlier — and ask how the change in their time moves with the change in race-morning heat.

To get the per-distance figures at the top, we collect every such pair — about 199,000 of them — and measure how the time gap moves with the heat gap. The slope is the "slower per +5°C" number. Restricting a runner to the very same course gives the lower end of each range; using all their races at that distance gives the upper end.

These figures rest on a lot of repeat runners: about 89,000 at 10K, 34,000 at the half and 21,000 at 5K. The marathon, with only around 2,500, is the least certain.

We also checked it a second way, comparing how a whole field's typical finish shifted between cooler and warmer editions instead of following individuals. It points the same way at each distance, so the result isn't an artefact of how we link runners across races.

The full within-athlete method, and how we decide two results belong to the same runner, are on the methods page ↗.

The heat measure

"Hotter" here isn't a thermometer at the finish line; few races record one. We use a modelled reconstruction of each race morning, roughly 4am to noon, for the race's city and date. It is humidity-aware: a muggy 24°C morning counts as hotter than a dry 24°C, because the body sheds heat less easily when the air is already damp.

It is also a shade estimate: it has no direct-sun term. Standing in the sun on race morning is harder than these numbers show, so they read as conservative, not inflated. More on the weather data ↗

What it doesn't capture

Only finishers

The data only sees people who finished. Anyone who dropped out on a brutal day is invisible, so the heat cost here is a floor. The runners hurt worst aren't in the sample dragging the average down further.

Shade, not sun

The heat measure has no direct-sun component, so on a clear, exposed course the real cost is likely steeper than shown.

The season rides along

Comparing a runner with themselves controls who you are, not when you race. A pair that crosses from a cool-season race to a warm-season one also crosses training cycles and race goals, and that inflates the upper end of each range. Comparing only races in similar months shrinks the numbers — at 10K toward the low end of the range, at the half below it — though the effect is also genuinely steeper the hotter the morning already is. Read the ranges as what a warmer morning costs in practice, weather plus everything that travels with it.

Marathons are thin

Few runners repeat a marathon, and when they do it's often on a different course, so the marathon figure is the least certain and is shown as a range.

For how the data is gathered and cleaned, how runners are identified, and the fitness-drift caveat in full, see Methods & data ↗.