Effects of time-restricted eating on body composition and performance in athletes: a systematic review and meta-analysis
A systematic review and meta-analysis of how time-restricted eating (TRE) affects athletes' energy intake, body composition and performance. Randomized controlled trials in adult athletes that set TRE against non-TRE eating patterns were sought in six databases from inception to January 10, 2026; 245 athletes across sixteen studies from 15 articles were included.
Energy intake did not change significantly with TRE (SMDpooled = −0.35, with 95%CI: −0.77 to 0.07 and p = 0.11).
Body mass fell slightly but significantly (SMDpooled = −0.27, with 95%CI: −0.50 to −0.04 and p = 0.02), as did body fat% (SMDpooled = −0.39, with 95%CI: −0.68 to −0.11 and p = 0.01); fat mass and lean mass did not change, and aerobic capacity, muscular endurance, muscular strength and power were not significantly affected. Certainty of evidence was rated very low to moderate.
The authors conclude TRE may be a workable way for athletes to reduce body mass and body fat% modestly without clear harm to performance, while noting that small samples, short interventions and methodological concerns limit the evidence.
Participants
Sixteen studies from 15 articles, 245 athletes, adult (inclusion criterion: adult athletes).
202 men and 43 women, mean ages 18.4 to 39.9 years; resistance-trained athletes, runners, taekwondo athletes and DanceSport athletes (full text, checked 2026-10-02: section 3.1). The reference list of included trials also contains the elite-cyclist trial already in the base as REF-520.
Protocol
Search: six databases from inception to January 10, 2026; PRISMA; PROSPERO CRD420261291351.
Included: randomized controlled trials in adult athletes that tested TRE against a non-TRE eating pattern and reported either body composition or performance.
Pooling: standardized mean differences (SMDs; Hedges' g) with 95% confidence intervals, random-effects models; risk of bias by RoB 2, certainty by GRADE.
TRE as applied (full text, checked 2026-10-02: sections 2.2 and 3.2): defined as an eating window of 8-12 h; most trials used 16/8, one used 12/12; windows generally began between 10:00 and 13:00 and ended between 18:00 and 21:00. Controls mostly ate their habitual diet on a comparable training programme.
Energy intake handling differed between trials (full text, checked 2026-10-02: section 3.2): three imposed no caloric restriction or nutrition intervention, two monitored intake and compensated for deficits, one adjusted intake weekly to promote mass gain, one prescribed a 25% caloric deficit in both groups, and the rest changed meal timing on habitual or self-selected diets.
Duration 4 to 52 weeks, most 4-8 weeks; effect sizes were computed from post-test means and SDs (full text, checked 2026-10-02: sections 3.3 and 2.3).
Key Results
Energy intake: not significantly altered (SMDpooled = −0.35, 95%CI: −0.77 to 0.07, p = 0.11).
Body mass: small but significant reduction (SMDpooled = −0.27, 95%CI: −0.50 to −0.04; p = 0.02).
Body fat%: small but significant reduction (SMDpooled = −0.39, 95%CI: −0.68 to −0.11, p = 0.01).
Fat mass: unchanged (SMDpooled = −0.23, 95%CI: −0.48 to 0.02, p = 0.07).
Lean mass: unchanged (SMDpooled = −0.16, 95%CI: −0.42 to 0.09, p = 0.21).
Aerobic capacity: no significant effect (SMDpooled = 0.04, 95%CI: −0.29 to 0.37, p = 0.82).
Muscular endurance: no significant effect (SMDpooled = −0.20, 95%CI: −0.52 to 0.13, p = 0.23).
Muscular strength: no significant effect (SMDpooled = 0.01, 95%CI: −0.25 to 0.27, p = 0.94).
Power: no significant effect (SMDpooled = −0.12, 95%CI: −0.44 to 0.20, p = 0.47).
In subgroup analyses, elite athletes may have lost more body mass and body fat%, but the differences between subgroups were not statistically significant.
Certainty of evidence ranged from very low to moderate.
Certainty by outcome (full text, checked 2026-10-02: sections 3.6.1-3.6.3): moderate for body fat%; low for energy intake, body mass, fat mass, lean mass, aerobic capacity, muscular endurance, muscular strength and power; very low for protein and carbohydrate intake.
The "lean mass unchanged" result is not robust (full text, checked 2026-10-02: section 3.6.2.4): in the sensitivity analysis that accounts for correlated estimates the lean-mass loss excluded zero (SMDpooled = −0.14, 95%CI: −0.27 to −0.01, p = 0.03), and two individual trials reported a significant lean-mass reduction. The same sensitivity analysis made the fat-mass reduction significant (SMDpooled = −0.22, 95%CI: −0.32 to −0.12; section 3.6.2.2).
Protein intake pointed lower on TRE without reaching significance and with high heterogeneity (SMDpooled = −0.82, 95%CI: −1.87 to 0.23, p = 0.13; I² = 85.92%) (full text, checked 2026-10-02: section 3.6.1.2).
Aerobic capacity came from five studies and was primarily VO2max and VO2peak; no time trial or race result was pooled (full text, checked 2026-10-02: section 3.4).
Risk of bias: eleven studies some concerns, four high risk, none low (full text, checked 2026-10-02: section 3.5).
Practical Application
**"Can I use an 8-hour eating window to lean out without getting slower or weaker?" In a meta-analysis of 16 small randomized trials (245 athletes), time-restricted eating gave small reductions in body mass (SMD −0.27, 95%CI −0.50 to −0.04) and body fat% (SMD −0.39, 95%CI −0.68 to −0.11), with lean mass (SMD −0.16, 95%CI −0.42 to 0.09) and aerobic capacity, muscular endurance, strength and power not significantly changed; energy intake was not significantly lower (SMD −0.35, 95%CI −0.77 to 0.07).The authors rate the certainty very low to moderate.** Small, and not shown to cost performance.
Who and how long: 202 of the 245 were men, most trials used a 16/8 pattern and most lasted 4-8 weeks (full text, checked 2026-10-02: sections 3.1-3.3) — a short-term result in mostly male, mostly resistance-trained athletes.
The honest reading of the mechanism: the intake estimates all point downward (energy, protein, carbohydrate) though none is significant, and several trials did not control or report intake. The authors themselves say timing cannot be separated from energy balance. Inference, not tested here: the window probably works to the extent that it makes the athlete eat less.
Lean mass deserves the caveat, not the headline: the main model says "unchanged", the sensitivity model says a small loss.
Not from this paper; the base's own practice: the low-energy-availability screen comes before any advice on eating less or eating in a window (SYNTH-54, built on REF-341). This meta-analysis pooled no hormone, bone, menstrual or other REDs indicator, so it cannot clear TRE on that count; the one trial in the base that measured hormones found free testosterone and IGF-1 falling on an 8-h window (REF-520).
What it does not show: anything about a season-long window, endurance race performance, women in numbers (43 participants), or masters athletes.
Limitations
It is a real meta-analysis (inverse-variance random-effects pooling of Hedges' g, RoB 2 — the right tool for randomized trials — and GRADE, PROSPERO-registered), but of weak trials: no study at low risk of bias, 245 participants in total, most interventions 4-8 weeks (full text, checked 2026-10-02: sections 2.4, 3.5, 5).
Timing is confounded with intake. Several trials did not control or report energy and macronutrient intake; one put both groups in a 25% deficit, one fed a surplus. The authors name this as a limitation (full text, checked 2026-10-02: sections 3.2 and 5). The pooled energy-intake estimate rests on a handful of studies — the text says six and cites five.
Results depend on how correlated estimates are handled: fat mass and lean mass are "unchanged" in the main model and significantly lower in the robust-variance sensitivity analysis; the authors acknowledge the sensitivity (full text, checked 2026-10-02: sections 3.6.2.2, 3.6.2.4, 5). The abstract reports only the main model.
Effect sizes were computed from post-test values, not changes from baseline (full text, checked 2026-10-02: section 2.3), so with groups this small a baseline difference between groups passes straight into the estimate; crossover trials are among the included studies and how they were handled is not stated.
"Athletes" is broad: Tier 1-2 recreational exercisers are pooled with Tier 3-5, and the titles of some included trials describe "healthy men" or "young males" (full text, checked 2026-10-02: section 2.4 and reference list). Mostly resistance-trained men; 43 women (section 3.1).
Performance means laboratory tests — VO2max, repetitions to failure, 1RM, jumps; no pooled time-trial or race outcome, and each performance outcome comes from four to six studies (full text, checked 2026-10-02: section 3.4).
No health outcomes pooled: no hormones, bone, menstrual function, mood or eating-disorder measures, so the review is silent on low energy availability.
Heterogeneity was I² = 0.00% for every body-composition and performance outcome (full text, checked 2026-10-02: sections 3.6.2-3.6.3), which with trials this small says the trials were too imprecise to disagree, not that they agree. Egger's tests showed no publication bias where ten or more studies allowed the test.
Selection: 37 full texts were excluded as "no full text" and 2 were unavailable; non-English articles were excluded (full text, checked 2026-10-02: sections 2.2 and 3). The eating-window definition differs inside the paper (4-12 h in the introduction, 8-12 h in the methods).
The elite-versus-recreational subgroup difference is not statistically significant.
Funding: none received; no conflict of interest declared (full text, checked 2026-10-02: Funding, Conflict of interest). No PMID in Europe PMC at the time of writing.
How this entry was checked and changed
Why this level
lowered from A-: no included trial was at low risk of bias (11 some concerns, 4 high), the authors' GRADE is low for body mass, lean mass, energy intake and every performance outcome (moderate only for body fat%), trials are small and mostly 4-8 weeks, and energy intake was not controlled, so eating window and eating less cannot be separated