REF-350 · Grgic et al., 2020 + Guest et al., 2021 (ISSN position) · British Journal of Sports Medicine, · Paper
Caffeine — umbrella review of 21 meta-analyses (Grgic 2020) plus the ISSN position stand (Guest 2021)
Key Findings (evidence summary)
11 systematic reviews containing 21 meta-analyses, searched across 12 databases; all of moderate or high methodological quality (AMSTAR 2)
Caffeine was ergogenic for aerobic endurance, muscle strength, muscle endurance, power, jumping performance and exercise speed — but not every analysis gave a definite direction once the 95% prediction interval was considered
Quality of evidence (GRADE) generally moderate, with some low to very low; muscle endurance, muscle strength, anaerobic power and aerobic endurance were the outcomes with moderate-quality support
The effect is generally greater for aerobic than for anaerobic exercise
Most primary studies were in young men
Dosing and timing — ISSN position stand (Guest 2021)
Performance improves consistently at 3-6 mg/kg body mass; the minimal effective dose is unclear but may be as low as 2 mg/kg
Very high doses (e.g. 9 mg/kg) bring a high incidence of side-effects and are not needed for an ergogenic effect
Most common timing: 60 min pre-exercise; optimal timing depends on the source — caffeinated chewing gum may need a shorter lead time than capsules
Effective in both trained and untrained individuals; aerobic endurance shows the most consistent moderate-to-large benefit
Inter-individual differences in performance response and in side-effects (sleep, anxiety) may be due to genetic variation in caffeine metabolism and response; habitual intake may also play a role
Endurance exercise in the heat: 3-6 mg/kg supported; at altitude: 4-6 mg/kg
Alternative forms (gum, mouth rinse, gels, chews) improve performance, mainly aerobic; energy drinks and caffeinated pre-workouts enhance anaerobic and aerobic performance
Caffeine also improves attention and vigilance, and may help physical and cognitive performance under sleep deprivation
Practical Application
Twenty-one meta-analyses agree caffeine is ergogenic across endurance, strength, power, jumping and speed, with the largest and most consistent gains in aerobic endurance; the ISSN puts the working dose at 3-6 mg/kg about 60 min before exercise, possibly effective from 2 mg/kg. For a race or key session, that is the evidence-backed window — higher doses (~9 mg/kg) add side-effects, not performance.
Start at the low end (2-3 mg/kg) and test in training; genetics and habitual intake shape both the benefit and the sleep/anxiety cost, and the ISSN notes the individual response varies.
Gum acts faster than capsules — take it closer to the start; mouth rinse is covered separately in REF-422.
Not from this paper; the base's own practice: keep caffeine away from the hours before sleep (SYNTH-07 for the timing rationale) and avoid untested high-dose pre-workout cocktails on race day.
Limitations
Most primary studies in young men; women, middle-aged and older adults under-studied (Grgic 2020's own call)
Not all meta-analyses gave a definite direction once prediction intervals were considered; some outcomes rest on low to very low quality evidence
Genetic and habitual-use effects are acknowledged but not quantified in either abstract; no genotype-specific dose can be given from these sources
How this entry was checked and changed
Content correction 2026-09-27
Effect sizes "ES ~0.20-0.40 endurance, ~0.10-0.25 sprint, ~0.10-0.20 strength" -> not in either abstract; removed (the umbrella review reports direction and evidence quality, not pooled ES).
"30-60 min pre", "peak plasma 45-60 min", "split dosing 2-3 mg/kg at 90-120 min", "7-10 day abstinence", CYP1A2 −163 A/C frequencies (~50%, ~10-15%) and genotype dose ranges, ADORA2A, "6-8 h before sleep, half-life 5-6 h", "gum ~15 min" -> none in either abstract; removed. Dosing now carries only the ISSN abstract's figures (3-6 mg/kg, ≥2 mg/kg, 9 mg/kg, 60 min, heat 3-6, altitude 4-6).
Sport "running, swimming, sprint" -> general (broad range of exercise tasks).
2026-09-28: added Journal line for the DOI paper (Grgic 2020) from Europe PMC (PMID 30926628).