The caffeine dose that measurably improves performance is 3 to 6 milligrams per kilogram of body mass, taken about sixty minutes before exercise. For a 70 kg person that works out at 210 to 420 mg, or roughly two to four cups of filter coffee. That figure comes from the official International Society of Sports Nutrition position stand on coffee, published in 2023.

Most of the rest of what gets said about coffee and sport holds up less well. The thermogenic effect will not make you leaner. Coffee’s antioxidants change nothing measurable about recovery. And the advice most often repeated to vegetarians, keep coffee an hour away from an iron-rich meal, only works in one direction: an hour before, the effect disappears; an hour afterwards, iron inhibition is exactly the same as if you had drunk it with the food.

What caffeine actually does, sport by sport

The effect is real, and it is smaller than gym culture suggests. More importantly, it varies a lot depending on what you measure.

In swimming, a 2026 multilevel meta-analysis pooled thirteen randomised placebo-controlled crossover trials covering 144 men and 48 women, contributing 28 performance effect sizes. Caffeine improved performance with a standardised effect size of 0.57, which translates to a 1.71% gain on the clock. On a 100 metre swim taking a minute, that is about one second. For a club swimmer, one second is a lot. For someone doing lengths on a Tuesday evening, it is invisible.

Exploratory analyses in the same review suggested doses of 6 mg/kg or above produced larger pooled effects than lower doses. The authors flag this as hypothesis-generating and in need of confirmation, which is a careful way of saying you should not turn it into a dosing rule.

For strength, the numbers are considerably more modest. A three-level meta-analysis published in 2026 included 24 studies in female participants, 23 of them in the quantitative synthesis, contributing 107 effect sizes. Maximal strength gained g = 0.17, with a confidence interval of 0.06 to 0.29 and low GRADE certainty. Muscular endurance showed a larger mean effect, g = 0.54, but with a prediction interval running from 0.06 to 1.02, meaning the result reproduces poorly from one protocol to the next. Meta-regressions on dose, age and resistance load identified no reliable moderation.

Across all performance outcomes in women, a separate 2026 meta-analysis covering twenty studies and 144 primary effect sizes found g = 0.37, confidence interval 0.24 to 0.50. The prediction interval crosses zero. Put plainly: on average it works, but in a randomly chosen future study the effect could be nil.

A 2026 trial using a balanced-placebo design in sixteen resistance-trained men at 5 mg/kg shows the variability well. Caffeine and the expectation of caffeine raised peak torque by 4%, but only 50 to 88% of participants counted as responders depending on the measure, and improvement among those responders ranged from 1 to 16%.

Dose, timing and form

The ISSN position stand specifies two to four cups, roughly 473 to 946 ml of typical hot-brewed or reconstituted instant coffee, depending on individual sensitivity and body size, sixty minutes before exercise. That volume delivers the caffeine equivalent of 3 to 6 mg/kg, along with 100 to 400 mg of chlorogenic acids per cup.

The vagueness is not editorial laziness. Coffee is a complex matrix of hundreds of compounds, consumed with broad variability depending on serving size, bean type (arabica versus robusta) and brew method: water temperature, roasting, grind size, time and equipment. Two espressos pulled in two cafés on the same street do not deliver the same dose. This is the most awkward practical limitation in the whole topic, and the position stand acknowledges it directly, noting a dearth of literature examining coffee-specific ergogenic and recovery effects, plus variability in the operational definition of the word “coffee” between studies.

A 2026 narrative review compared coffee and isolated caffeine head to head as ergogenic aids. Isolated caffeine consistently improves performance under controlled conditions. Coffee produces comparable effects, particularly in endurance exercise, but responses are more variable, precisely because of differences in caffeine content and individual sensitivity. Coffee, on the other hand, appears better tolerated by many people, and the authors conclude that the two should not be treated as interchangeable: isolated caffeine remains the experimental tool, coffee remains the realistic everyday vehicle.

Iron, and the advice that needs fixing

This is where precision matters most, because the mistake is common and it has real consequences once meat is off the plate.

The reference study, published in 1983 in the American Journal of Clinical Nutrition, used dual isotope methodology in iron-replete subjects. A cup of coffee taken with a meal reduced non-heme iron absorption by 39%. Tea, a recognised inhibitor, managed 64%. With a meal made from semipurified ingredients, absorption fell from 5.88% to 1.64% with drip coffee and to 0.97% with instant. Doubling the strength of the instant coffee dropped it to 0.53%.

Then comes the finding most pages leave out. No decrease in iron absorption occurred when coffee was consumed one hour before the meal. But when it was taken one hour afterwards, inhibition matched what was seen with simultaneous ingestion. The after-lunch coffee, the one you drink without thinking about it, costs as much iron as the one you drink with the food.

For someone eating meat, this stays a secondary question: heme iron from meat is absorbed by a pathway coffee does not block. For someone whose dietary iron is almost entirely non-heme, the effect applies to the whole intake. It is the only mechanism in this topic that genuinely shifts with what you eat, and it is fixed by timing rather than by quantity.

If you train, iron is not a detail: it carries oxygen. The strategies that raise absorption still apply, and we cover them in our article on optimising plant-based iron absorption. Vitamin C with the meal, soaking your legumes, spacing out the inhibitors. Coffee simply joins the list of inhibitors, with an asymmetric schedule: before is harmless, during or after is costly.

The real trade-off: performance against sleep

The 3 to 6 mg/kg recommendation has a blind spot when the session or the competition happens in the evening. A 2025 review in Sports Medicine was devoted entirely to this conflict: the performance literature recommends taking caffeine, the athlete sleep literature advises against the same decision, especially before evening competition.

Figures from the EFSA scientific opinion on the safety of caffeine give the scale of the problem. Single doses of 100 mg, about 1.5 mg/kg for a 70 kg adult, may increase sleep latency and reduce sleep duration in some individuals, particularly when consumed close to bedtime. Elsewhere the opinion notes that single doses of 1.4 mg/kg and above taken at bedtime could impair sleep in some people.

In other words: the dose that helps performance is two to four times the dose that is enough to damage a night. Population data confirm this at scale. An analysis of 37,469 adults from NHANES 2005-2018 found an adjusted odds ratio of 1.5 for short sleep above 400 mg per day, against 1.1 between 100 and 399 mg and 1.0 below 100 mg. Intakes of 101 to 400 mg taken within eight hours of bedtime were associated with short sleep.

For a morning session the question does not arise. For an 8 pm match, taking 4 mg/kg at 7 pm means betting better performance against a degraded night and the recovery that follows. The review authors recommend resolving this individually, using simulated competitions with dual measurement: performance during exercise and unwanted effects in the hours afterwards. If sleep is already your weak point, our article on tryptophan and sleep tackles the other end of the problem.

Safety thresholds, in numbers

The EFSA opinion sets two reference points for healthy adults. Single doses up to 200 mg, about 3 mg/kg for a 70 kg adult, from all sources, do not raise safety concerns, even if consumed less than two hours before intense physical exercise under normal environmental conditions. And intakes up to 400 mg per day, about 5.7 mg/kg, do not raise safety concerns for the general adult population.

Two caveats matter here. First, the opinion states that no studies are available in pregnant women or in middle-aged and elderly subjects undertaking intense physical exercise. Second, it notes that single doses of 3 mg/kg and above could increase anxiety in some cases.

For pregnant women, EFSA identifies 200 mg per day from all sources as not raising safety concerns. For lactating women, single doses up to 200 mg and intakes up to 400 mg per day do not raise safety concerns for the breastfed infant.

The friction point is easy to see: the upper ergogenic dose recommended by the ISSN, 6 mg/kg, exceeds EFSA’s single-dose reference of 3 mg/kg. This is not two incompetent institutions contradicting each other. The ISSN describes a dose producing a measurable laboratory effect in athletes; EFSA describes a threshold below which it flags no concern for the general population. Taking 6 mg/kg before a race is a deliberate performance choice, not a habit to install four times a week.

What the anti-doping status actually says

Caffeine is not on the prohibited list. It appears on the World Anti-Doping Agency’s 2026 Monitoring Program, under stimulants, in-competition only, alongside bupropion, nicotine, phenylephrine, phenylpropanolamine, pipradrol and synephrine.

The document explains what that category means: the World Anti-Doping Code, at Article 4.5, provides that WADA shall establish a monitoring program covering substances that are not on the Prohibited List but that it wishes to monitor in order to detect potential patterns of misuse in sport. Drinking coffee before a competition therefore carries no sanction. The substance is simply being watched.

What a vegetarian diet actually changes about performance

The question underneath this page deserves a direct answer, because it circulates widely in the form of anxiety.

A randomised crossover trial published in 2026 in Nutrition Journal followed 36 recreational athletes from four universities, sixteen runners and twenty resistance trainers. Each completed two four-week diets, omnivore and plant-based, in random order, with meals served in university dining halls. The mean difference on the 12-minute timed run was -36.2 metres, confidence interval -130.3 to +57.9 metres, p = 0.43. On composite machine strength the difference was -4.0%, interval -8.5% to +0.6%, p = 0.08. Neither reached significance.

Among female athletes, a 2026 narrative review reaches the same place by a different route: available intervention and observational studies generally report comparable performance, physiological and health outcomes between vegetarian and omnivorous diets. The authors make a useful point: much of the variability observed across studies reflects implementation factors (energy availability, risk of low energy availability, protein adequacy, micronutrient intake, dietary quality, duration of adherence) rather than dietary classification itself.

The practical conclusion is less dramatic than either side of the debate: the diet is not the limiting factor, the execution is. Eat enough, cover the micronutrients, sustain it over time. Our pages on vegetarian diets and endurance sport and on building muscle on a vegetarian diet cover that execution.

Creatine: where being vegetarian does change something measurable

The one parameter where vegetarian status produces a clear gap is not iron or protein, it is muscle creatine. It comes from meat and from endogenous synthesis, and vegetarians start with lower stores.

A randomised, double-blind, placebo-controlled trial published in 2025 followed fifteen healthy vegans and vegetarians for seven days on 0.3 g of creatine monohydrate per kilo per day, split across four daily doses, against a maltodextrin placebo. The creatine group gained 1.56 kg of body mass and 1.15 kg of fat-free mass. Muscle creatine rose by 18.8 mmol/kg and total muscle creatine by 30.8 mmol/kg.

And performance? Unchanged. Phosphocreatine stayed level, and neither peak nor mean power output during repeated sprints moved. Filling the tank did not make the car go faster over this seven-day protocol with fifteen participants. That is worth knowing before buying a tub of powder on the strength of a marketing claim.

Does coffee really upset digestion during exercise?

This worry comes up constantly in running and cycling forums, and it deserves more precision than “some people are sensitive to it”.

A preliminary study published in 2025 in Physiological Reports tested exactly this. Eighteen men and women ingested 3 mg/kg of caffeine or placebo 45 minutes before cycling: twenty minutes at 70% of maximal oxygen uptake, followed by a fifteen-minute time trial. Researchers measured plasma intestinal fatty acid binding protein, a marker of damage to intestinal epithelial cells.

Endurance exercise damages gut cells, which is established. What this study found is that caffeine worsens that damage, particularly in participants classified as sensitive. The effect was influenced by ADORA2A genotype rather than CYP1A2 genotype. The authors describe their work as preliminary and call for further examination of the implications for gastrointestinal responses to exercise. It is a signal rather than a certainty, but it points somewhere uncomfortable for anyone downing a double espresso before a long ride.

Genetics: what it explains and what it does not

CYP1A2, the gene coding the enzyme that metabolises most caffeine, has become the standard explanation for differences between individuals. Reality is more qualified.

A 2025 systematic review included nineteen randomised clinical trials published between 2012 and 2024, totalling 732 participants, with doses from 2 to 9 mg/kg. Results point the same way without reaching consensus: positive effects in AA genotype carriers, positive but smaller in AC carriers, null or negative in CC carriers. An important caveat rides along with that result: 84% of included studies were rated “some concerns” in the risk of bias analysis, with only two rated low risk.

A triple-blind randomised trial published in 2026 tested 94 resistance-trained participants (47 women, 47 men; 39 AA, 44 AC, 11 CC) at 3 mg/kg. Caffeine improved strength and muscular endurance across conditions. Mean velocity rose by 4 to 12% in AA and 3 to 9% in AC individuals, against changes of 4% or less in CC carriers. But the supplement by genotype interaction appeared only at 90% of one-repetition maximum. The authors conclude that genotype did not alter the overall effect and explains only part of individual variability.

Getting genotyped to decide whether to have a coffee before a session is therefore a gadget. Testing on yourself, and recording honestly what happens, is more informative and costs nothing.

Adolescents: why the answer is different

If you coach a young athlete, the reasoning does not transfer.

A 2026 systematic review with three-level meta-analysis included 35 studies in athletes aged 10 to 19, of which 32 contributed to the quantitative analysis with 726 participants and 363 effect sizes. The primary analysis found a small favourable average effect on physical performance, g = 0.33, confidence interval 0.17 to 0.49. Pooled effects were not statistically significant for sport-specific performance, perceptual response or physiology. Exploratory analyses provided no reliable evidence that effects differed by dose or ingestion timing.

The authors are explicit: current evidence is insufficient to support routine caffeine use in youth sport. No serious adverse events were reported, but adverse-event assessment was inconsistent between studies.

EFSA, for its part, applies the reference derived from acute adult consumption, 3 mg/kg per day, to children and adolescents, given the limited information specific to that group. It also notes that about 8% of adolescents aged 10 to under 18 may consume more than 200 mg of caffeine from energy drinks in a single session connected with physical exercise.

Claims worth dropping

Three ideas circulate widely and do not survive scrutiny.

The thermogenic effect as a body composition argument. Caffeine does slightly raise energy expenditure, but the magnitude does not justify treating it as a lever. No data support presenting coffee as a weight management tool for an athlete.

Coffee antioxidants as recovery support. Coffee does contain substantial chlorogenic acids, 100 to 400 mg per cup according to the ISSN position stand. But the 2026 narrative review on coffee and physical performance is clear: evidence regarding the non-caffeine bioactive compounds in coffee remains limited and largely indirect, indicating a modulatory rather than ergogenic role. Your fruit and vegetables do that job better and without caffeine.

Coffee as a glycogen resynthesis accelerator. The hypothesis circulates, and it is testable. A double-blind crossover trial published in 2026 tested it in twelve trained cyclists using a whole coffee cherry extract delivering 200 mg caffeine and 15 mg polyphenols, with muscle biopsies at 0, 4 and 24 hours and 1 g of carbohydrate per kilo of body mass. The result: the extract was ergogenic in the time trial but did not enhance muscle glycogen resynthesis. Performance and glycogen recovery do not behave the same way.

How to test it on yourself

The methodological recommendation from the Sports Medicine review deserves repeating, because you can apply it without a laboratory: simulate the competition, with dual measurement. Performance during the effort on one side, unwanted effects in the hours afterwards on the other.

In practice that means holding everything else still. Same session, same time of day, same pre-session meal, and as far as possible the same sleep the night before. Then alternate two conditions across four to six comparable sessions, with and without caffeine, recording three things: the objective result (time, load, repetitions), how hard the effort felt, and the quality of the night that follows.

Three precautions make the test less misleading. Try the lower dose first, 3 mg/kg rather than 6, because the gap in side effects between the two is wider than the gap in performance. Do not run the test on a competition day, where adrenaline masks everything. And accept that the answer may be “no measurable difference”: only 50 to 88% of participants counted as responders in the balanced-placebo trial, which leaves a substantial minority in whom the effect simply does not show up.

Keep the iron question separate while you do this. If you move your coffee earlier to protect a meal, you have changed two variables at once, and the session comparison stops meaning anything. Sort the timing question first, then test the dose.

If the following night degrades reliably, the question is no longer whether caffeine improves your session. It becomes what that improvement costs across the full training week.

Common mistakes

Drinking your coffee at the end of the meal because you read you should “space it by an hour”. The spacing only protects in one direction. An hour before the meal, iron absorption is unaffected. An hour after, it is hit as hard as during.

Using the same dose regardless of session time. At 7 am, 4 mg/kg has no consequence for the night ahead. At 7 pm, the same dose falls inside the eight-hour pre-bedtime window associated with short sleep.

Counting in cups without knowing what is in them. The ISSN position stand stresses variability by bean, grind, temperature and method. Two cups can mean 120 mg or 300 mg.

Looking for the effect in an espresso swallowed five minutes before the start. The documented window is sixty minutes before exercise.

Raising the dose because the effect has faded. The position stand notes that coffee’s physiological effects are influenced by habituation to a small degree. Increasing the dose mostly shifts the side-effect dial, notably anxiety above 3 mg/kg according to EFSA.

Treating coffee as equivalent to water during exercise. Water and sports drinks remain the base, as set out in our article on hydration and electrolytes for vegetarian athletes.

By profile

You run or cycle in the morning, with no known iron issue. Pre-session coffee is the best-documented use there is. Allow sixty minutes, stay between 3 and 6 mg/kg, and keep your iron-rich meals outside the coffee window.

You train late in the day. This is where the trade-off is real. Test without caffeine first, or with the lower dose. If you choose caffeine before an evening event, understand that you may be paying in sleep, and that sleep is the main recovery tool you have.

You are vegetarian and your iron stores run low. The useful lever is timing, not abstinence. Coffee away from main meals, never at the end of one, and watch out for tea, which inhibits more. Our article on plant-based iron absorption covers the other levers, and the one on coffee and acidity handles the digestive side.

You lift. The documented effect on maximal strength is small, g = 0.17 in women, with low certainty. Not nothing, but not a major lever either. Training volume and protein intake weigh more, which we cover in our complete guide to plant proteins.

You get digestive symptoms during exercise. The gut cell study suggests caffeine worsens exercise-induced damage in sensitive individuals. One session without caffeine, compared against one with, will tell you more than any article.

You coach under-18s. The evidence does not support routine use. EFSA’s 3 mg/kg per day reference is the frame, and energy drinks are the real problem before coffee is.

You are pregnant or breastfeeding. EFSA’s reference is 200 mg per day from all sources during pregnancy, and the opinion notes the absence of studies in pregnant women doing intense exercise. Discuss it with your doctor rather than with an article.

Limits of what we know

The isotope study on iron dates from 1983 and involves small numbers of iron-replete subjects. Its figures are precise, its base is narrow.

The ISSN position stand acknowledges a dearth of literature examining coffee-specific ergogenic and recovery effects, as opposed to caffeine in its other delivery forms, as well as variability in the operational definition of “coffee”.

Recent meta-analyses in women report low to very low GRADE certainty, with prediction intervals crossing zero for several outcomes.

The crossover trial on plant-based diets and performance covers 36 participants recruited at four universities, over four weeks per diet. That is short for capturing adaptation.

The creatine trial in vegetarians covers fifteen people and seven days. The absence of a performance effect there is a result, not a definitive refutation.

The gut cell study is explicitly described as preliminary by its authors.

The NHANES sleep analysis is cross-sectional and relies on 24-hour dietary recall: it establishes associations, not causation.

Frequently asked questions

How many cups before a race? The ISSN position stand indicates two to four cups of hot-brewed or reconstituted instant coffee, roughly 473 to 946 ml, sixty minutes beforehand, depending on sensitivity and body size.

Is decaf just as useful? Not for performance: the documented effect comes from caffeine. It keeps its place if you want the taste without the stimulation, particularly in the evening.

Should I stop coffee for a few days before a competition to “reset” sensitivity? The ISSN position stand notes habituation influences effects to a small degree. A 2026 structured review on problematic use and withdrawal in athletes found only four eligible studies and concludes that current evidence is insufficient for firm conclusions. Pre-competition withdrawal remains a hypothesis, not an established protocol.

Is espresso more effective than filter? Nothing indicates so for performance. What matters is the total caffeine dose and the timing, not the extraction method.

Does coffee count towards my daily fluid intake? It contributes to daily fluid intake. It does not replace water during exercise.

I drink coffee and eat vegetarian: should I take an iron supplement? That decision belongs with a blood test and a healthcare professional, not with an article. Coffee influences dietary iron absorption; it does not diagnose a deficiency.

Coffee before or after training? Before, if the goal is performance. Afterwards, the only thing the data document is that a coffee extract with carbohydrate did not improve glycogen resynthesis.

Is caffeine banned in competition? No. It sits on WADA’s 2026 Monitoring Program, stimulants category, in-competition only, meaning it is observed rather than prohibited.

Sources

International Society of Sports Nutrition position stand on coffee and sports performance, 2023. Comparative review of coffee versus isolated caffeine, Nutrients 2026. Narrative review on coffee and physical performance, Nutrients 2026. Review of the caffeine and sleep conflict in athletes, Sports Medicine 2025. 2026 meta-analyses on swimming, on female performance and on resistance exercise in women. Systematic review and randomised trial on CYP1A2 genotype, 2025-2026. 2026 meta-analysis in adolescent athletes. 2025 study on caffeine and exercise-induced gut cell damage. 2026 trial on coffee cherry extract and glycogen resynthesis. 2025 trial on creatine in vegans and vegetarians. 2026 randomised crossover trial on plant-based diets and performance. 2026 review on vegetarian diets in female athletes. Dual isotope study on the inhibition of food iron absorption by coffee, American Journal of Clinical Nutrition 1983. EFSA scientific opinion on the safety of caffeine. World Anti-Doping Agency 2026 Monitoring Program. Swiss dietary recommendations for adults from the FSVO and the WHO healthy diet fact sheet.