Answers are grounded in this paper and cite their sources.
Background Caffeine is a well-established aid for endurance performance under normoxic conditions. More recently, its ergogenic potential to improve endurance performance in hypoxia has attracted increasing interest, although findings remain inconsistent. This meta-analysis therefore evaluated the effects of caffeine on endurance performance under hypoxic conditions.Methods PubMed, Web of Science, Scopus, and the Cochrane Library were searched up to December 2025 for placebo-controlled, randomized crossover studies examining the impact of caffeine on endurance performance (e.g. time-trial and time-to-exhaustion) and physiological and/or metabolic responses (e.g. blood lactate concentration, heart rate, arterial oxygen saturation, and minute ventilation) in hypoxic conditions. Risk of bias was assessed with the Cochrane Risk of Bias tool. Meta-analyses were conducted in R software, comparing caffeine with placebo, with pooled effects calculated as Hedge's g using random-effects models.Results Nine studies were included (n = 117; female: n = 19). Caffeine significantly improved exercise performance in hypoxia, increasing exercise capacity across both time-to-exhaustion and time-trial tests (g = 0.76, 95% CI = 0.34 to 1.18, p < 0.001, low GRADE). Caffeine also increased heart rate, blood lactate concentration, and minute ventilation (g = 0.40 to 0.47, all p = 0.005, low GRADE). No significant effects were observed for ratings of perceived exertion or arterial oxygen saturation (g = -0.25 to 0.17, all p > 0.05, very low to low GRADE) under hypoxic conditions.Conclusion This meta-analysis indicates that caffeine enhances endurance performance in hypoxic conditions and is accompanied by changes in key physiological and metabolic responses. Consequently, caffeine supplementation may serve as a practical strategy for attenuating hypoxia-induced performance impairments in athletes training or competing at altitude. Future high-quality, rigorously designed studies are warranted to further validate these effects.
No takes yet. Share an insight, caveat, or question.
Peng et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: