REF-778 · Bossi et al., 2017 · International Journal of Sports Physiology and Performance, Vol. · Paper · PubMed

Pacing Strategy During 24-Hour Ultramarathon-Distance Running

B+moderate evidenceMeasured on recreational to highly trained athletesWritten before October 2026, being re-checked

Hourly running speeds of 398 men and 103 women across 5 editions of a 24-h race (all with at least 19.2 h of effective running) were normalised to each runner's 24-h mean speed. Runners covered 135.6 ± 33.0 km with 22.4 ± 1.3 h of effective running. Overall, pace followed a rough reverse J-shape, with a significant drop in speed from the second-to-last to the last hour.

Pacing differed by performance group and by race edition but not by sex or age group. Better performers started at lower relative speeds: performance was inversely associated with normalised speed in the first 2 h (r = -.58) but not in the last 2 h (r = .03).

Participants

398 male and 103 female participants of 5 editions of a 24-h race, included if they reached a minimum of 19.2 h of effective running.

Protocol

Mean running speed for each hour normalised to the runner's 24-h mean speed.
Two-way mixed ANOVAs: racing time × sex, age group, performance group and race edition.
Pearson correlations between performance and normalised speed in the first and the last 2 h.

Key Results

Performance: 135.6 ± 33.0 km; effective running time 22.4 ± 1.3 h.
Overall pattern: reverse J-shaped pacing, with a significant reduction in speed from the second-to-last to the last hour.
Performance group × racing time: significant interaction (F = 7.01, P < .001, ηp2 = .04).
Race edition × racing time: significant interaction (F = 3.01, P < .001, ηp2 = .02).
Sex × racing time: not significant (F = 1.57, P = .058). Age group × racing time: not significant (F = 1.25, P = .053).
Start pace vs result: performance vs normalised speed in the first 2 h, r = -.58 (P < .001); in the last 2 h, r = .03 (P = .480).

Practical Application

In 398 men and 103 women over 5 editions of a 24-h race, the better performers started at lower speeds relative to their own 24-h average and paced more evenly (performance vs first-2-h relative speed r = -.58), while the last 2 h did not separate good from poor results (r = .03) (observational). In a 24-h race, the opening hours are where a result is lost: start at, not above, the pace you can hold for the day.
Sex and age group did not change the pacing pattern, so the same start-conservatively advice applies to women and men and to older and younger runners in this dataset.
The pattern shifted between race editions; the authors read this as "herd behaviour". Inference, not tested here: the pack around a runner at the start pulls them faster, so decide the first hours' pace before the gun.
The speed drop in the final hour is part of the typical shape here; the paper does not say why. Inference, not tested here: it may reflect stopping or walking before the clock ends as much as fatigue.
Not from this paper; the base's own practice: pace a timed ultra by effort and a pre-set speed ceiling for the first hours, and plan fuelling and breaks as part of pacing.

Limitations

Observational, one race over 5 editions; course, climate and field size belong to that race.
Only runners with at least 19.2 h of effective running were included, so those who stopped early (and may have paced worst) are excluded.
Speeds are hourly means; walking, stops and sleep inside each hour are not separated.
Association only: starting slower went with better results, but the study cannot show that starting slower causes them.
Written from the abstract and record metadata only (full text not read).
How this entry was checked and changed
Why this level

large observational dataset, 398 men and 103 women; single race