The current record, Dennis Kimetto’s 2:02:57, was noticeably more even in pace than previous records.
The current record, Dennis Kimetto’s 2:02:57, was noticeably more even in pace than previous records.
Sweat Science

What’s the Best Way to Pace a Marathon?

An analysis of five decades of world records reveals a surprising shift in pacing approach

The current record, Dennis Kimetto’s 2:02:57, was noticeably more even in pace than previous records.

Get full access to Outside Learn, our online education hub featuring in-depth fitness, nutrition, and adventure courses and more than 2,000 instructional videos when you sign up for Outside+.

One of the big questions in the buildup to Nike’s Breaking2 marathon last year was the pacing strategy. To run a two-hour marathon, should you plan to run at exactly two-hour pace for the entire race? Should you start a little quicker to bank some time against the threat of late-race slowdown (a fast-slow approach known as “positive splits”)? Or should you hold back at the beginning to feel good for as long as possible, then use the excitement of the approaching finish to accelerate (a slow-fast approach known as “negative splits”)?

In theory, you can construct pretty reasonable physiological arguments for all three approaches. Several months before the Breaking2 race, when I asked Nike’s scientific team about their pacing plans, they still hadn’t decided. Part of the challenge, they explained, is that pacing isn’t just about physiology. They needed to ensure that the three athletes they’d selected were fully comfortable and confident about whatever strategy they selected—and at that point, the athletes were evenly split. One wanted positive pacing, another wanted negative pacing, and the third wanted even pacing.

In other words, the “right” way to pace a marathon is complicated. One way to gain insights into what works best is to get outside the laboratory and study how the fastest marathons in history have been run. That’s what researchers in Spain, led by senior author Jordan Santos-Concejero of the University of the Basque Country, have done in a new study published in the European Journal of Sport Science.

Drawing on data collected by the Association of Road Running Statisticians, Santos-Concejero and his colleagues analyzed the pacing patterns of the most recent 15 men’s world records in the marathon, dating back to Derek Clayton’s 2:09:36 in Fukuoka in 1967. They divided the race into eight 5K sections, plus a final 2.195K finishing section. Overall, the athletes tended to run the second half of the race slightly (pretty much negligibly) faster than the first half.

Here’s what the first-half and second-half speeds look like, as a percentage of overall race speed:

(European Journal of Sport Science)

On the surface, this looks like a pretty good endorsement of the even-split school of pacing. But there’s a catch. When the researchers split the records into two groups, the records prior to 1988 showed a distinctively different pacing pattern compared to the records since then.

Here’s what the splits from the two groups look like over the nine sections of the race (eight 5K sections plus 2.195K):

(European Journal of Sport Science)

The older “classic” records are characterized by a fast start, a progressive slowdown after about 25K, then a last gasp of reacceleration in the final 2.195K (though even that finishing kick is slower than their overall average race speed). The newer “contemporaneous” records do exactly the opposite, starting more slowly than the eventual average pace and accelerating after 25K.

So, what’s going on here? The authors point out that the break point between the two eras happens to coincide with Ethiopian runner Belayneh Dinsamo’s 1988 record. Of the older records, three were set by Australians, one by a Welshman, and one by a Portuguese runner. Of the newer records, the tally is four by Kenyans, three by Ethiopians, two by a Moroccan-born runner, and one by a Brazilian runner.

Do East African runners pace themselves differently? That’s a fraught question, but the authors point out previous research (some by Santos-Concejero) suggesting that Kenyan elite runners are better able to maintain oxygen levels in their brains during exhaustive exercise compared to runners of European origin. Perhaps, they speculate, there’s some link between brain oxygenation and the ability to accelerate during the second half of a marathon. Interestingly, Ethiopian runner Abebe Bikila, who set a pair of marathon world records in the early 1960s (prior to the period analyzed in this study), paced himself more like the “new” group of record setters.

The other big question is whether the shift to negative splits is the best way of running a fast marathon. In another subanalysis, the authors calculate the “coefficient of variation” of the nine subsegments of each race, which is a measure of how even or uneven the splits were. Plotting that number over the years, there appears to be a slight trend toward smaller variations, meaning that runners now tend to keep their pace within a narrower band for the entire race. The best example of this is the current record, Dennis Kimetto’s 2:02:57, which was noticeably more even than previous records.

With this in mind, the authors suggest that “a pacing strategy characterized by very little speed changes across the whole race may be the way to go in the future.” To be honest, I think the evidence for this statement is weak (the coefficient of variation data isn’t very convincing), but I’m inclined to suspect that it’s correct, thanks to Occam’s razor, if nothing else. That, in the end, is what Nike’s Breaking2 team opted for, instructing its pacing team to run at precisely two-hour pace for as long as possible.

In the real world, of course, the challenge with perfectly even pacing is that you have to know exactly how fast you plan to run before you even start. It requires hindsight, and perhaps some circular logic, to conclude after the fact that the best way to run a given time is to have started out at exactly that pace. If we knew exactly what our capabilities were before every race, then trying to run that pace as evenly as possible would be a no-brainer.

But when you add the uncertainties inherent in real life, with time flowing in the forward direction, you still have to decide whether you’re going to err on the side of caution or ambition in your early pacing. And I’m not sure physiology will ever provide a definitive answer to that, because the “right” approach depends on your goals and how you weight them. Is it 3:10 or bust for you? Then you should start out at a 3:10 pace. But if 3:15 or 3:20 are still meaningful secondary goals, then perhaps starting at a 3:13 pace maximizes your overall chance of a positive outcome even if it makes it slightly less likely that you’ll hit 3:10.

I do think Santos-Concejero and his colleagues are probably correct that for future elite runners to keep whittling away at the world record, they’ll have to set out at world record pace right from the start and maintain as even a pace as possible. The consequence, for those runners good enough to dream of records, will be lots of spectacular flameouts—which can be fun to watch but are less fun to participate in. For the rest of us, aiming for a slight negative split still sounds like a pretty good plan.

My new book, Endure: Mind, Body, and the Curiously Elastic Limits of Human Performance, with a foreword by Malcolm Gladwell, is now available! For more, join me on Twitter and Facebook, and sign up for the Sweat Science email newsletter.

When you buy something using the retail links in our stories, we may earn a small commission. We do not accept money for editorial gear reviews. Read more about our policy.

promo logo