REF-1673 · Zheng et al., 2026 · Frontiers in Physiology, · Paper · PubMed

In team ball sport athletes after a presumed normal night, an early-afternoon nap improved intermittent "endurance" tests, mostly 5-m shuttle runs (SMD 0.96, 95% CI 0.62 to 1.29), and agility (0.55, 0.01 to 1.09) against no nap, while power and strength (0.50, −0.09 to 1.09), cognitive and technical results were not clearly changed, in a meta-analysis of 12 small randomised crossovers (179 athletes, 171 M / 8 F); naps over 60 min gave the largest endurance effect (1.38) on very few studies

Main caveat: The authors': few studies per domain, low power and no assessment of publication bias; small samples that may inflate the endurance effect.
fairly strong evidenceMeasured on trained to highly trained athletesRead in full

Zheng and colleagues asked how a daytime nap changes objective performance in team ball sport athletes and whether nap length matters. They searched five databases to December 2025 and included 12 randomised crossover trials with 179 athletes (soccer, rugby, handball, basketball, volleyball), all comparing a nap with a no-nap control after a presumed normal night.

Naps started between 13:00 and 15:30, most often at 14:00, and lasted 20-90 min. Outcomes were grouped into five domains and pooled with robust variance estimation.

The nap improved endurance-type tests by a large amount and agility by a moderate amount; the power-and-strength estimate was positive but its interval crossed zero, and the cognitive and technical estimates were too imprecise to read.

By duration, naps of 31-60 min and over 60 min both improved endurance, with the largest point estimate for long naps; no duration subgroup for power or agility had an interval above zero. The authors suggest integrating moderate or long naps into routines, caution that the effects may be inflated by small studies, and call their recommendations heuristics.

Participants

12 trials, 10-25 athletes each, 179 in total: 84 soccer, 29 rugby, 28 handball, 22 basketball, 16 volleyball players.
Mean ages 17.2-27.6 years by trial; all-male samples except one volleyball trial (8 M / 8 F).
Countries: Tunisia 5, France 2, Turkey 2, New Zealand, Japan, China 1 each. Five Tunisian trials share one laboratory, sports mix (football, rugby, handball) and age (about 20 years), so some may share participants (our reading; not addressed in the text).
Prior night: presumed normal; sleep checked by actigraphy in some trials, by self-report or questionnaire in others. One trial had a simulated late-evening match the night before.

Protocol

Design of included trials: randomised crossover, washout 2-7 days; nap start 13:00-15:30; durations 20-90 min; control = quiet wakefulness or awake rest.
Domains: cognitive and attention; agility; power and strength; endurance (Yo-Yo test, total distance, most often the 5-m shuttle run test); technical skill.
Analysis: SMDs from post-intervention means (within-person correlations rarely reported), time-based outcomes sign-reversed so positive = better; robust variance estimation (robumeta, clubSandwich, ρ = 0.7; sensitivity at 0.2, 0.5, 0.8); subgroups short (≤ 30 min), moderate (31-60 min), long (> 60 min); no meta-regression (fewer than 10 studies per domain).
Quality: RoB 2; no GRADE; publication bias not assessed (too few studies).

Key Results

Domain, nap vs no napSMD (95% CI)p; dfτ²
Endurance (intermittent tests)0.96 (0.62 to 1.29)< 0.001; 4.440
Agility0.55 (0.01 to 1.09)0.048; 2.930
Power and strength0.50 (−0.09 to 1.09)0.078; 3.710
Cognitive and attention0.88 (−1.97 to 3.73)0.390; 2.861.42
Technical skill (2 studies)0.91 (−9.21 to 11.03)0.457; 1.000.93
Subgroup by nap durationSMD (95% CI)p; df
Endurance, 31-60 min0.85 (0.46 to 1.23)0.004; 3.55
Endurance, > 60 min1.38 (1.18 to 1.58)0.002; 1.83
Power, ≤ 30 min0.58 (interval not given; crosses zero)—
Power, 31-60 min0.29 (interval not given; crosses zero)—
Power, > 60 min (2 studies)0.84 (−0.44 to 2.13)0.076; 1.00
Agility, ≤ 30 min0.57 (−0.10 to 1.23)0.072; 2.94
Agility, 31-60 min0.55 (−0.25 to 1.34)0.096; 1.93
Changing the assumed within-study correlation moved no pooled SMD by 0.01 or more.
The pre-specified sensitivity analysis without the trial that had an evening match the night before, and the leave-one-out results, are described as planned but their results are not reported in the text.
Risk of bias: 3 trials (25%) some concerns, 1 (8.3%) high.

Practical Application

For a team-sport player after a normal night, an early-afternoon nap of 30-90 min before a later session the same day is a reasonable bet for repeated-effort running and agility: shuttle-type endurance tests improved by about one standard deviation (SMD 0.96) and agility by about half (0.55, an interval that only just clears zero); jump and strength, decision and skill measures were not clearly changed.
Agrees with [SYNTH-07]'s nap window: naps in these trials started 13:00-15:30, inside the guide's 20-90 min before 16:00. Naps over 60 min gave the largest endurance effect (1.38), but on so few studies (df 1.83) that it does not support preferring long naps over 30-60 min ones (0.85).
Wake-up buffer — the authors differ from the guide, without new data: they suggest at least 60 min between waking and competition, citing one earlier study, while SYNTH-07 uses 30 min awake before training (REF-608); this review did not test the interval, so the guide's figure stands and 60 min is a cautious option before a match.
Inference, not tested here: for cyclists, runners and triathletes the "endurance" here is not their kind (5-m shuttles, Yo-Yo, match distance); transfer to a continuous endurance session is untested, and in a 600-km race napping did not go with better results (REF-854, self-selected).

Limitations

The authors': few studies per domain, low power and no assessment of publication bias; small samples that may inflate the endurance effect; chronotype and habitual napping not reported or controlled; nap sleep often self-reported; mostly young male athletes.
Analysis: crossover trials analysed from post-intervention values as if the groups were independent, which ignores the pairing; agility (p = 0.048) and the long-nap subgroups rest on two or three effective degrees of freedom or fewer.
Possible shared samples: five Tunisian trials from one laboratory with similar athletes (our reading).
Reporting: the abstract presents the long-nap power estimate (0.84) among improvements, though its interval crosses zero; the study table marks one agility result of SMD 0.08-0.09 as clearly positive, and gives cognitive SMDs of 3.73-5.85 for one trial; the planned sensitivity analyses are not reported; no GRADE.
No funding; no conflicts of interest declared.
How this entry was checked and changed
Why this level

risk of bias read in the full text (3 of 12 trials some concerns, 1 high risk, so 8 low by our arithmetic), but the trials are all small (10-25 athletes), each domain rests on few studies (effective df 1.0-4.4), crossover data were pooled as if from independent groups, publication bias could not be assessed and no certainty grading was done