Caffeine and Sport-Specific Endurance Performance (Time-Trial Studies): Systematic Review
Scope
Ganio and colleagues restricted the review to studies that measured endurance performance with a time-trial component (set distance or time, best effort) lasting more than 5 minutes — arguing that time-to-exhaustion tests are too variable and do not correspond to how any endurance sport is won. 21 studies with 33 identifiable caffeine treatments qualified; methodological quality was high (mean PEDro 9.3 out of 10).
Key Findings
Mean improvement in time-trial performance with caffeine: 3.2 ± 4.3%
Range across studies: −0.3% to +17.3% — highly variable; the authors attribute the spread to ingestion timing, mode/vehicle and subject habituation
Effective dose: moderate quantities, 3-6 mg/kg body mass, taken before and/or during exercise
Habituation: abstaining from caffeine for at least 7 days before use gives the greatest chance of the full ergogenic effect
Practical Application
Systematic review of 21 time-trial studies (Ganio 2009): caffeine at 3-6 mg/kg taken before and/or during exercise improved endurance time-trial performance by 3.2 ± 4.3% on average, but the response ranged from −0.3% to +17.3% between studies — so the dose is settled, the individual response is not.
Test your own response in training time trials before relying on it in a race: the between-study range says some athletes gain little.
If you want the largest effect, the review's advice is a caffeine break of at least 7 days beforehand.
Not from this paper; the base's own practice: start at 3 mg/kg for an untested athlete, dose 45-60 min before the start, and check sleep and anxiety at that dose — dosing detail and genetics in REF-350.
Limitations
Systematic review with a computed mean, not a weighted meta-analysis; the 3.2% figure carries a standard deviation larger than itself.
Time-trial studies only (>5 min); nothing on sprint or strength outcomes.
Timing, vehicle (coffee, capsule, gel) and habituation were not resolved by the review — it names them as the likely sources of variability.
Published 2009; the genetics (CYP1A2) and mouth-rinse literature came later.
How this entry was checked and changed
Content correction 2026-09-27
"Meta-analysis" -> systematic review of 21 studies / 33 treatments with a computed mean (no pooled weighted estimate); title and Type corrected.
"+3-4% (medium effect)" -> 3.2 ± 4.3%, range −0.3 to +17.3%; "3-6 mg/kg 30-60 min before exercise" -> 3-6 mg/kg before and/or during exercise (timing window not in the abstract); "7-10 days withdrawal" -> at least 7 days.
"80-85 kg example 250-490 mg", "race day 3-5 mg/kg (250-410 mg) 60 min before" -> house arithmetic and timing, not from this paper; removed (base practice kept as a labelled line pointing to REF-350).
Cross-reference descriptions checked against the cited entries' current content; wrong years, wrong entries and claims the cited entry does not make were corrected.