All articles
6 min readrunning / endurance

Even pacers hit the wall 0.2% of the time: how to pace a marathon

I race 100 km, and the stretch I plan hardest is the first hour, when nothing hurts yet. Among 78,912 Boston Marathon finishers, the runners who held an even pace fell apart late in 0.2% of cases. The runners with erratic splits fell apart in 85.4% of cases. Those two numbers come from the same race, the same weather and the same course.

Even splits and a 59-minute gap.

Qizheng Dong and Zaidong Li clustered the nine split times of 78,912 Boston Marathon finishers from 2015 to 2017 1. Four patterns came out of the data: even pacing (47.9% of the field), mild positive splitting (31.9%), strong positive splitting (12.2%) and variable pacing (8.0%). Mean finish time was 217 minutes in the even group against 276 minutes in the strong positive-split group. The authors counted a pace drop above 20% in the final segment as hitting the wall. It caught 15.2% of the whole field, 0.2% of the even-pacing cluster and 85.4% of the variable-pacing cluster. Even pacing also grew rarer as finish times grew slower.

Causation runs in both directions here. A 2:50 runner has the fitness to hold pace, so tidy splits are partly a symptom of being fit rather than a cause of it. Keep that in mind for the lab trials further down.

Fast starters are the ones who collapse.

Barry Smyth measured the start pace of 1.7 million recreational marathon runners as a percentage of each runner's own mean race pace 2. The quickest finish times sat at or very near that pace, and the finish-time cost climbed on both sides of it. Fast starters carried a far higher risk of late-race collapse than cautious starters, who mostly lost a few minutes to timidity. Women in his data paced closer to their mean than men did.

In a later analysis of that collapse, Smyth reported that 28% of male runners hit the wall against 17% of women 3.

Robert Deaner and colleagues had already measured the same sex gap in slowing across 14 US marathons. Mean change in pace from the first half to the second was 15.6% for men and 11.7% for women (p < 0.0001), and the gap showed up in all 14 races 4. Runners with more marathons behind them slowed less, and controlling for that experience left the sex difference standing.

Over 5 km, the fast start is free.

Amy Gosztyla and co-authors had 11 moderately trained women distance runners race 5 km after two baseline trials, with the first 1.63 km set at their baseline average pace, 3% faster, or 6% faster 5. Finish times came in at 21:11 ± 29 s for the even start, 20:52 ± 36 s at 3% and 20:39 ± 29 s at 6%. The three times did not differ reliably (p > 0.05). The 6% start produced the fastest trial for 8 of the 11 runners, and the 3% start for the other 3.

Push the distance to 20 km and the sign flips. Craig Mattern and colleagues had 13 competitive male cyclists ride a self-paced 20 km time trial, then two more trials where minutes 1 to 4 were held 15% below and 15% above their own self-selected starting power 6. The slow start beat the fast start (p < 0.05), and blood lactate at minute 4 was lower on the slow start. After minute 4, ventilation, heart rate and perceived effort matched.

The pooled evidence sets a ceiling on all of this. Domingo Ramos-Campo and colleagues ran a network meta-analysis over 24 pacing studies, 302 participants and 765 performance observations 7. No imposed pacing strategy outperformed letting the athlete choose their own. Drawn-out slow starts impaired performance. Peak oxygen uptake, mean oxygen uptake and end-test lactate did not separate the strategies.

Those lab trials last 10 to 60 minutes, and almost nothing runs out in an hour. The marathon findings come from four-hour races where glycogen, core temperature and muscle damage have time to pile up. So the split discipline below belongs to races past about two hours. Over 5 km you can open hard and get away with it.

The protocol.

  1. Derive mean race pace from a race you have finished. Add 5% to your most recent half marathon's per-kilometre pace. A 1:45 half (4:59/km) plans out at 5:14/km, or 3:41 for the marathon.
  2. Write the 5 km split down and treat it as a floor on elapsed time. At 5:14/km that is 26:10. Plan the first 5 km at plan pace or 2 seconds per kilometre slower, the region where Smyth's finish-time cost is lowest.
  3. Audit once at 5 km and hand back any gain over the next 5 km. 30 seconds under plan at the first checkpoint means running the second 5 km 6 seconds per kilometre slower. In my own races this is the correction that matters most, and it is over by kilometre 10.
  4. Band all eight 5 km splits, plus the final 2.2 km, at ±2% of plan. At 5:14/km that is 5:08 to 5:20 per kilometre. Print the eight cumulative times on a wristband so you read a target instead of doing arithmetic at 25 km.
  5. Cap your second-half drift at 5%. Deaner's field slowed 15.6% and 11.7%. A 5% cap on a 5:14/km plan lets the last 10 km run at 5:30/km, which costs 2 minutes 40 seconds against plan.
  6. Drop the whole rule under 25 minutes of racing. For a 5 km, open the first mile 3 to 6% faster than your average pace, as the 11 runners in Gosztyla's trial did.
  7. Rehearse the exact split, not the feeling. In the four weeks before a marathon I run the first 5 km of two long runs at the planned split and nothing faster. Race pace feels too easy at 5 km, and that is the sensation to recognize under start-line adrenaline.

Before your next long run, work out your 5 km target from step 1 and write the number on your hand. Look at the watch once, at 5 km, and see how far off you were.

Sources.

  1. Dong, Q., Li, Z. (2026). Pacing strategy patterns and performance outcomes in marathon running: A large-scale analysis of split time data. PLOS ONE, 21(7), e0352808. doi:10.1371/journal.pone.0352808
  2. Smyth, B. (2018). Fast starters and slow finishers: A large-scale data analysis of pacing at the beginning and end of the marathon for recreational runners. Journal of Sports Analytics, 4(3), 229-242. doi:10.3233/JSA-170205
  3. Smyth, B. (2021). How recreational marathon runners hit the wall: A large-scale data analysis of late-race pacing collapse in the marathon. PLOS ONE, 16(5), e0251513. doi:10.1371/journal.pone.0251513
  4. Deaner, R.O., Carter, R.E., Joyner, M.J., Hunter, S.K. (2015). Men are more likely than women to slow in the marathon. Medicine & Science in Sports & Exercise, 47(3), 607-616. doi:10.1249/MSS.0000000000000432
  5. Gosztyla, A.E., Edwards, D.G., Quinn, T.J., Kenefick, R.W. (2006). The impact of different pacing strategies on five-kilometer running time trial performance. Journal of Strength and Conditioning Research, 20(4), 882-886. doi:10.1519/R-19275.1
  6. Mattern, C.O., Kenefick, R.W., Kertzer, R., Quinn, T.J. (2001). Impact of starting strategy on cycling performance. International Journal of Sports Medicine, 22(5), 350-355. doi:10.1055/s-2001-15644
  7. Ramos-Campo, D.J., Foster, C., Andreu-Caravaca, L., Rubio-Arias, J.Á., Rosenblat, M.A. (2026). Comparative Effects of Pacing Strategies on Endurance Performance: A Systematic Review and Network Meta-Analysis. Sports Medicine, 56(3), 725-738. doi:10.1007/s40279-025-02367-3

Keep reading