Caffeine for performance and alertness
Does caffeine actually improve exercise performance and mental sharpness?
Yes. Now the interesting part.
Caffeine works. It is one of a very small number of compounds we cover where the answer to "does this do anything" is an unqualified yes, backed by a literature deep enough that the arguments have moved on to second-order questions.
Which makes it the most useful compound on this site for a different reason: it shows what a genuinely settled question looks like, so you have something to compare the unsettled ones against.
The size of the evidence base
Most compounds we grade have a meta-analysis. Caffeine has a review of the meta-analyses.
A 2020 umbrella review searched twelve databases and pooled eleven systematic reviews containing 21 separate meta-analyses, every one assessed as moderate or high methodological quality.2 Its finding was that caffeine was ergogenic — performance-enhancing — for aerobic endurance, muscle strength, muscle endurance, power, jumping performance and exercise speed.
Synthesis of the currently available meta-analyses suggest that caffeine ingestion improves exercise performance in a broad range of exercise tasks... It seems that the magnitude of the effect of caffeine is generally greater for aerobic as compared with anaerobic exercise. Grgic et al., British Journal of Sports Medicine, 20202
The International Society of Sports Nutrition reached the same place in its 2021 position stand, whose first conclusion describes "small to moderate benefits" across muscular endurance, movement velocity, strength, sprinting, jumping and throwing, and whose second singles out aerobic endurance as "the form of exercise with the most consistent moderate-to-large benefits."1
The dose that works, and the dose that doesn't
This is where caffeine stops being trivia and becomes practical — and where a genuine tension appears that almost nobody mentions.
What the literature and the regulators each say
Read the last three rows together. The dose most consistently shown to improve performance sits at or above the single-dose amount European regulators describe as raising no safety concerns. The bottom of the ergogenic range lines up with EFSA's single-dose figure; the top of it is roughly double.
That is not a warning that the ergogenic dose is dangerous — EFSA's figures are thresholds below which there is no concern, not thresholds above which there is harm. But it does mean the performance literature and the safety literature are describing different amounts, and anyone reading only one of them gets an incomplete picture.
On timing, the position stand notes that 60 minutes pre-exercise is most commonly used, that optimal timing depends on the source, and that chewing gum may need a shorter wait than capsules.1 It also concludes that caffeine improves performance in trained and untrained people alike, and that habitual users do not reliably lose the benefit — study findings on tolerance "remain equivocal."1
Where our grade and the reviewers' grade disagree
Something worth being upfront about, because it looks like a contradiction and isn't.
We grade caffeine Strong. The umbrella review, applying the GRADE framework, rated the underlying quality of evidence as "generally moderate (with some low to very low quality of evidence)."2 Both are correct, because the two scales ask different questions.
Two scales, two questions
Trial count, participant totals, independent replication, meta-analytic pooling Strong
Risk of bias, imprecision, indirectness, publication bias within studies Moderate
A very large literature made of individually imperfect trials scores high on ours and middling on theirs. Neither reading is wrong, and a reader is better served knowing both than being handed one number. Where a formal GRADE assessment exists for a compound we cover, we will report it alongside our own mark rather than quietly preferring whichever is more flattering.
The reviewers also flag a real limitation on generalisability: most individual studies in the pooled meta-analyses were conducted among young men, and they call for more primary research in women, middle-aged and older adults.2 That caveat travels with every number on this page.
The genotype question, which is not settled
You will increasingly see the claim that caffeine's benefit depends on your CYP1A2 genotype — that "fast metabolisers" gain and "slow metabolisers" gain nothing or are actively impaired. Genetic testing companies sell against this idea.
The position stand does treat individual variation as real, attributing differences in response and in side effects to genetic variation in caffeine metabolism alongside factors such as habitual intake.1 So the general principle has support.
The specific exercise claim is another matter. A 2020 double-blind placebo-controlled crossover trial in 18 healthy adults at 3 mg/kg tested exactly this, and split the two outcomes apart:3
Carswell et al. 2020 — effect of genotype on caffeine response
Their conclusion: caffeine "enhanced CYP1A2 'fast' metabolisers' cognitive performance more than 'slow' metabolisers," but "no other between-genotype differences emerged for the effect of caffeine on exercise or cognitive performance, or metabolism."3
Eighteen people is small, and one trial does not overturn a field. But it points the other way from the confident version of this claim, and it separates two things that get sold together: a genotype effect on thinking found support here, while a genotype effect on cycling did not.
We have seen specific figures for genotype-related performance impairment quoted widely. We could not trace them consistently to a primary source — different secondary accounts report different numbers — so we are not repeating any of them. The honest summary is that this question is open.
Four claims, four answers
The cognitive claim sits a grade lower than the performance claims not because it is doubtful but because we have not located the same depth of pooled human evidence for it; the position stand states caffeine is ergogenic for attention and vigilance "in most individuals."1 If a comparable umbrella review of cognitive outcomes exists, that grade should rise.
Safety and unknowns
-
Adverse events reported
Very high doses — around 9 mg/kg — are associated with "a high incidence of side-effects" and confer no additional performance benefit.1 Effects on sleep and feelings of anxiety are documented, and vary between individuals partly for genetic reasons.1
-
Established intake thresholds
EFSA: single doses up to 200 mg (about 3 mg/kg) and daily intakes up to 400 mg (about 5.7 mg/kg) do not raise safety concerns for healthy adults. Single doses as low as 100 mg may affect sleep duration and patterns in some adults, particularly close to bedtime.4
-
Known interactions and at-risk groups
Pregnant and breastfeeding women: EFSA advises staying below 200 mg per day from all sources. Children and adolescents: 3 mg/kg per day.4 These are lower than the doses shown to improve performance.
-
Longest human exposure studied
Caffeine is among the most consumed substances on earth and has been studied for decades, but the performance trials pooled here are overwhelmingly acute — single doses before a single exercise bout.2 Long-term ergogenic use is a different question from long-term dietary consumption.
-
What has not been characterized
Performance effects in women, middle-aged and older adults, which the umbrella review names as the field's main generalisability gap.2 Whether genotype meaningfully modifies exercise response. Whether habitual high intake blunts the acute effect — findings remain equivocal.1
How this was graded
Evidence strength — Strong, criteria checked
This is what our top grade is supposed to look like: not one impressive trial, but a literature so replicated that the reviews have reviews. Two compounds on this site have reached it. That ratio is the point of the scale.
Bottom line
Caffeine improves exercise performance, most reliably aerobic endurance, at 3–6 mg/kg taken about an hour beforehand. This holds for trained and untrained people, and habitual coffee drinkers do not clearly lose the benefit. It also improves attention and vigilance. There is no serious argument about any of this.
Three things temper it. The effective dose sits at or above the single-dose amount European regulators treat as unremarkable. The evidence is overwhelmingly from young men. And the popular claim that your genotype determines whether it works for you is not settled — the one trial we found that separated cognition from cycling supported it for the first and not the second.
A compound can be genuinely effective and still be oversold in the details. Caffeine is the cleanest example on this site of both halves being true at once.
Sources
- Guest NS, VanDusseldorp TA, Nelson MT, Grgic J, Schoenfeld BJ, Jenkins NDM, Arent SM, Antonio J, Stout JR, Trexler ET, Smith-Ryan AE, Goldstein ER, Kalman DS, Campbell BI. International society of sports nutrition position stand: caffeine and exercise performance. Journal of the International Society of Sports Nutrition. 2021;18:1. PMC7777221
- Grgic J, Grgic I, Pickering C, Schoenfeld BJ, Bishop DJ, Pedisic Z. Wake up and smell the coffee: caffeine supplementation and exercise performance — an umbrella review of 21 published meta-analyses. British Journal of Sports Medicine. 2020;54(11):681–688. PMID 30926628
- Carswell AT, Howland K, Martinez-Gonzalez B, Baron P, Davison G. The effect of caffeine on cognitive performance is influenced by CYP1A2 but not ADORA2A genotype, yet neither genotype affects exercise performance in healthy adults. European Journal of Applied Physiology. 2020;120(7):1495–1508. PMC7295849
- European Food Safety Authority. Caffeine. efsa.europa.eu
