REF-854 · Knechtle et al., 2012 · Chinese Journal of Physiology, Vol. · Paper · PubMed

No improvement in race performance by naps in male ultra-endurance cyclists in a 600-km ultra-cycling race

Blimited evidenceMeasured on trained athletesWritten before October 2026, being re-checked

Sixty-seven recreational men starting the 600 km Swiss Cycling Marathon (a Paris-Brest-Paris qualifier) were measured for anthropometry, training and race support. Finishers and non-finishers did not differ in anthropometry or training.

Among finishers, the riders who slept during the race were much slower than those who did not, and more sleep time went with a longer race time; after multivariate analysis only percent body fat and the duration of each training session remained predictors of race time. Support during the race and equipment showed no association with race time.

Participants

n=67 male recreational ultra-endurance cyclists at the 'Swiss Cycling Marathon' over 600 kilometres: 54 finishers and 13 non-finishers. Age, training level and body size are not given in the abstract.

Protocol

Anthropometry and training recorded; support during the race (riding alone vs followed by a support crew) and sleep time during the race recorded.
Bivariate analysis against overall race time, then multivariate analysis.
Finishers split into non-sleepers and sleepers for comparison of race time.

Key Results

Finishers vs non-finishers: no differences in anthropometry and no difference in training.
Support: finishers were significantly more often racing alone than followed by a support crew; no variable of support during the race was associated with race time.
Bivariate correlates of overall race time: percent body fat (r = 0.43), cycling distance per training unit (r = -0.36), duration per training unit (r = -0.31) and sleep time during the race (r = 0.50).
Sleepers vs non-sleepers (finishers):
GroupRace time, mean (IQR), min
23 non-sleepers1,567 (1,453-1,606)
31 sleepers1,934 (1,615-2,033)

P = 0.0003.

Multivariate: percent body fat (P = 0.026) and duration per training unit (P = 0.005) remained predictor variables for race time. Sleep time is not among the variables the abstract lists as remaining.

Practical Application

In 67 recreational men racing a 600 km ultra-cycling event, the 31 finishers who napped took 1,934 min against 1,567 min for the 23 who rode through (P = 0.0003), but in the multivariate model only percent body fat and duration per training unit remained predictors of race time. To "should I nap in a 600 km race?": in this race the nappers were clearly slower, yet the design cannot tell whether the naps cost the time or the slower riders were the ones who needed to sleep.
Inference, not tested here: a rider aiming for a fast time over a distance of about a day can plan to ride through; a rider who is struggling and dangerously sleepy should still sleep. The paper measured race time only, not safety, and the title's "no improvement" means naps did not make riders faster, not that sleep is unsafe to take.
Longer training sessions went with faster race times (duration per training unit stayed in the model). Inference, not tested here: for this kind of event the long ride is the training variable to protect; the same group's 2009 study of this race found no association between total logged training volume and race speed.
Riding without a support crew did not cost time here; finishers more often rode alone.
Not from this paper; the base's own practice: over multi-day races the evidence points the other way — riders who slept more per day ranked higher in the Race Across France REF-796; sleep planning for ultra events is in SYNTH-63.

Limitations

Observational: sleeping was self-selected, so the sleeper vs non-sleeper difference is not a causal effect of napping.
Men only, one race; age, training level and sleep durations are not given in the abstract.
The multivariate model's full variable list and effect sizes are not in the abstract.
The Crossref record lists no authors; first author confirmed from the PubMed and Europe PMC listings.
Written from the abstract and record metadata only (full text not read).
How this entry was checked and changed
Why this level

observational and correlational, n=67 in one race; sleeping was chosen by the riders, not assigned