Read the full writeup →
steady early pacethe wall

Data story · 31,655 marathon finishers

Hitting the Wall

Marathoners talk about "the wall" near 20 miles. Using every runner's 5-kilometre splits, I traced exactly where the field slows, and how hard.

By Jeremy Lee

~30 km
where pace starts falling off a cliff
+15%
median late-race slowdown vs mid-race pace
27%
of runners slowed more than 25% late on

What "the wall" actually looks like

Every finisher of the 2014 Boston Marathon crossed timing mats every 5 km. Converting those to a running pace and averaging across the field, a very clear shape appears: pace holds nearly flat for the first 20 km, drifts up gently to 25 km, and then climbs steeply from about 30 km onward. The textbook "wall." By 35–40 km the median runner is moving roughly a minute per kilometre slower than they were at the start.

Median pace by distance, rising sharply after 30 km
Median pace through the race. Flat early, then a steep climb (slowdown) past ~30 km.

It hits slower runners far harder

The wall isn't democratic. Measured against each group's own early pace, sub-3-hour runners barely slow at all, while runners finishing over 4:30 balloon by 30% or more in the closing kilometres. Fitness mostly shows up as the flatness of this curve.

Slowdown curves by finish-time group; slower groups rise much more
Slowdown relative to each group's own early pace. Faster groups stay flat; slower groups blow up late.

How bad is it, runner by runner?

Comparing each runner's 35–40 km pace to their steady 10–20 km pace, the median late slowdown is 15%, but the tail is long: more than a quarter of the field slowed by over 25% in the final stretch.

Histogram of late-race slowdown across runners
Distribution of late-race slowdown. A long right tail: the people who truly hit the wall.

The findings

The takeaway

"The wall" isn't a metaphor. It's a measurable bend in the pace curve that arrives, on average, right around 30 km. And because the bend is so much gentler for fitter, more even-paced runners, the practical lesson lines up with every coach's advice: train your endurance, and start conservatively, because the last 12 km is where marathons are won and lost. (See also my companion study, The Negative-Split Myth.)

Data & method: Official 2014 Boston Marathon results (public mirror: llimllib/bostonmarathon), 31,655 finishers with complete 5K splits. Each segment's pace = segment time ÷ segment distance. Reproduce with fetch_data.py + analyze.py in the repo. Boston's net-downhill start flatters early pace, so the real-world slowdown is, if anything, understated here.