Caffeine for Exercise: How Much by Body Weight, When, and What a Pre-Workout Scoop Gives You

How much caffeine do exercise studies use for your body weight?

Most studies use 3 to 6 mg of caffeine per kilogram of body weight, and some show benefits from as little as 2 mg/kg. The NIH Office of Dietary Supplements fact sheet on exercise describes many trials at about 2 to 6 mg/kg. The International Society of Sports Nutrition, in its 2021 position stand (the society's official summary of the evidence, with Nanci Guest as lead author), says 3 to 6 mg/kg works consistently, possibly as low as 2 mg/kg (Guest and colleagues, 2021^[1]).

To use the table, divide your weight in pounds by 2.2 to get kilograms, then read across.

Is more caffeine better?

A higher dose does not reliably give a bigger performance benefit. A meta-analysis pools the results of many studies into one average. One from 2022, covering 60 studies, found the same average benefit below 3 mg/kg as at 3 to 6 mg/kg; above 6 mg/kg the combined effect was smaller and could have been chance (Carvalho and colleagues, 2022^[2]). A 2025 analysis compared doses and forms across 48 time-trial studies (a time trial is a race against the clock over a set distance). It ranked capsules at 3 mg/kg or less first, but the ranking was not clear-cut: the likely benefit of 4 to 6 mg/kg capsules was nearly the same (Xue and colleagues, 2025^[3]).

One trial that found an advantage from a higher dose tested 13 resistance-trained men at 3, 6 and 9 mg/kg. Every dose sped up light lifts, but only 9 mg/kg improved the heaviest load (90% of the most they could lift once) and peak cycling power. At 9 mg/kg, side effects rose sharply, reported by 15% to 62% of the men depending on the symptom (Pallarés and colleagues, 2013^[4]). The 2021 position stand says doses around 9 mg/kg cause many side effects and do not seem to be needed. It lists a fast or pounding heartbeat, anxiety, headaches and poorer sleep as the most common problems, and says dose is the main thing that decides them.

Where a sports dose plus coffee crosses 400 mg

The FDA cites 400 mg a day, counted from all sources, as an amount not generally associated with negative effects in most healthy adults. That number is a daily safety benchmark, and a sports dose can reach it quickly. At 70 kg, 6 mg/kg is 420 mg on its own. At 80 kg, 3 mg/kg is 240 mg; add two 8-ounce cups of brewed coffee, which the NIH sheet puts at about 85 to 100 mg each, and the day is at 410 to 440 mg before any soda or tea. If you train with caffeine, the coffee you drank earlier counts toward the same total.

How long before training should you take caffeine, and does the form matter?

For capsules, tablets and coffee, most studies give caffeine about 60 minutes before exercise; the NIH sheet gives a range of 15 to 60 minutes. Gum is absorbed faster, so studies often give it closer to the start.

Capsules and tablets

The 60-minute habit rests on the assumption that blood caffeine peaks about an hour after you take it, and most trials follow it, though the gum study below measured the capsule peak later, at 84 to 120 minutes. A 2021 review put the best capsule timing at 30 to 60 minutes. One trial in women and men gave 3 mg/kg 90 minutes before a cycling test and still found a benefit, so the window is not tight. The position stand also notes that in long events, caffeine taken later (for example, cola near the end of a race) has worked about as well as a full dose an hour before.

Coffee

Coffee can work when it supplies the studied milligrams. In a small trial, eight trained male cyclists or triathletes took 5 mg/kg of caffeine either as coffee or as caffeine dissolved in water, one hour before exercise. Both finished a time trial about two minutes faster than placebo or decaf, with no statistically significant difference between the two caffeine drinks (Hodgson and colleagues, 2013^[5]). The older 2010 position stand had judged pure caffeine more effective than coffee; the 2021 stand allows coffee when the dose is high enough.

The practical catch is that you rarely know how much is in the cup. The 2021 stand cites an 8- to 9-fold spread in caffeine between coffee-shop servings. Most trials used capsules because the milligrams are fixed.

Gum and mouth rinse

Caffeine gum reaches peak blood levels sooner than a capsule, at about 44 to 80 minutes versus 84 to 120 minutes in one study of healthy men, with the same total amount absorbed. A 2026 meta-analysis of 25 gum studies (390 people) found a small but consistent benefit. Taking the gum closer to or further from the start made no clear difference, and in people who already used a lot of caffeine the benefit was too small to tell apart from chance (Miraftabi and colleagues, 2026^[6]). Direct comparisons now exist, and they do not identify one form that consistently performs better.

A 2025 trial gave 16 resistance-trained men 4 mg/kg as a capsule an hour before lifting or as gum chewed for five minutes before lifting, each compared with its own placebo. The most weight they could squat once rose by about 4.5% with capsules and 5% with gum. The difference between the active forms was not statistically significant. Capsules also improved the maximum bench press; the smaller gum result was uncertain. Gum still affected sleep: 7 of 16 reported insomnia after caffeine gum, versus 2 after placebo gum (Ding and colleagues, 2025^[7]).

The benefit also varies with the exercise. At 3 mg/kg, a study of 27 male rugby players found small combined caffeine effects on jump height and first-set squat repetitions, with most other tests unchanged; capsules, gum and mouth rinse did not differ clearly (Tallis and colleagues, 2024^[8]). A 15-man study found selected knee-strength and jump gains with 200 or 300 mg in capsules or gum, alongside other null strength results. It used only a placebo capsule, making the gum comparison harder to blind (Teimouri-Korani and colleagues, 2025^[9]). Neither form increased total CrossFit repetitions in a separate 13-man analysis: 6 mg/kg in a capsule before the 20-minute workout or gum during it was compared with a placebo drink (Vargas-Molina and colleagues, 2026^[10]).

Rinsing caffeine in your mouth and spitting it out has an even smaller effect. A 2026 meta-analysis of 31 studies found a trivial-to-small benefit, mostly in endurance exercise, with short rinses of about 5 seconds working better than longer ones (Deng and colleagues, 2026^[11]).

Morning or evening sessions

Several small trials in men found caffeine helped more in the morning, mainly by lifting morning performance up to evening levels. In 12 resistance-trained men, 3 mg/kg at 10 a.m. raised strength and power by about 5% to 6%, reaching the level they hit at 6 p.m. without caffeine (Mora-Rodríguez and colleagues, 2012^[12]). A second squat-speed study using 6 mg/kg and a study in teenage male basketball players also favored morning use (Mora-Rodríguez, 2015^[13]; Stojanović, 2022^[14]). The pattern was less clear in a 20-cyclist trial: performance improved at both times overall, although trained riders improved only in the morning (Boyett and colleagues, 2016^[15]).

Which kinds of exercise does caffeine help?

Caffeine helps endurance exercise most consistently, and it gives smaller benefits for strength, power and team sports. How much weight to put on the strength side is where official and sports-nutrition sources disagree.

Endurance

Endurance training is the best-studied use. A meta-analysis of 44 endurance time-trial studies at 3 to 6 mg/kg found average power up about 2.9% and finishing time down about 2.3% (Southward and colleagues, 2018^[16]; these figures come from the journal's published correction of the paper). On a one-hour time trial, 2.3% is about 80 seconds. Those are averages: some people improved more, and in some studies the result went the other way.

Strength and power

The NIH exercise sheet says caffeine is less likely to help sprinting or heavy lifting, relying on a 2010 position stand and a 2001 military report. The 2021 position stand reads the newer research differently and lists small-to-moderate benefits for strength, sprinting, jumping and throwing in many, but not all, studies.

The meta-analyses behind that update show real but small effects. A 2018 meta-analysis found a strength effect size of 0.20 for one-rep max (the most you can lift once) and 0.17 for jump power (Grgic and colleagues, 2018^[17]). An effect size compares the average change with how much people vary; around 0.2 is what researchers call small. Upper-body strength improved clearly in that analysis; the lower-body gain was too small to tell apart from chance. A 2020 review that pulled together 21 meta-analyses agreed that caffeine helps strength, power and speed, and that the effect is generally larger for aerobic exercise than for short, all-out efforts (Grgic and colleagues, 2020^[18]).

Lower-body findings have changed as more trials appeared. The 2018 analysis found an uncertain average gain, and the older 13-man trial found its heaviest-load benefit only at 9 mg/kg. The newer 16-man gum-versus-capsule trial described above found a higher maximum squat at 4 mg/kg with either form. That supports a moderate-dose benefit in those trained men.

Team sports

A 2019 meta-analysis of 34 team-sport studies at 3 to 6 mg/kg found small benefits across jumps, sprints, agility and running during matches. The largest effects were in total distance run in a match and the number of sprints, with effect sizes of about 0.4 (Salinero and colleagues, 2019^[19]). The authors call the effect small but unlikely to be chance.

Does a typical pre-workout scoop reach the studied dose?

For an average-sized adult, the typical labeled scoop lands near the low end of the studied range, but labels vary widely and many do not list the amount. The best US data are a count of label claims, not lab tests. In October 2018, researchers read the labels of the 100 top-selling pre-workout products on one large US retailer's site. Of those, 86 contained caffeine and 15 hid the amount in a proprietary blend. Among the products that listed a caffeine amount (71 of the 100), the average was 254 mg per serving. Forty-four products listed 210 mg or more (Jagim and colleagues, 2019^[20]).

Here is what 254 mg works out to by body weight:

The amount listed on the label may also not match what you're actually getting. An Australian lab tested 15 pre-workout products and measured 91 to 387 mg per serving. Only 6 of the 15 labels stated caffeine, and where they did, the measured amount ranged from 59% to 176% of the claim (Desbrow and colleagues, 2019^[21]).

If you're considering buying a pre-workout, find the caffeine milligrams per serving on the Supplement Facts label. Divide that amount by your weight in kilograms to compare the serving with the study doses above. Separately, add the milligrams from that serving, coffee and other sources to find your total for the day. A proprietary blend that lists caffeine without a number leaves you unable to do that math. Plant sources such as guarana also add caffeine; our guide to hidden caffeine in supplements goes over which other ingredients also include caffeine.

Does a daily coffee habit blunt the benefit?

Across the pooled research, regular coffee drinkers get about the same boost from a pre-exercise dose as people who rarely drink coffee. Two small trials that put non-users on daily caffeine, however, saw the benefit shrink within weeks. Both results stand, and they answer slightly different questions.

The pooled side: a 2022 meta-analysis of 60 studies found that how much caffeine people usually took in made no difference to the boost from a single dose before exercise. That held for endurance, power and strength exercise, for men and women, and for trained and untrained people (Carvalho and colleagues, 2022^[2]). In one trial within that literature, 40 male cyclists split by daily intake (about 58, 143 and 351 mg a day) all rode a 30-minute time trial faster on 6 mg/kg, with no difference between the groups (Gonçalves and colleagues, 2017^[22]).

The tolerance side: 18 people who normally took in little caffeine were given either daily caffeine (up to 3 mg/kg a day) or a placebo for four weeks. Before the four weeks, a 3 mg/kg dose boosted performance in both groups. After four weeks, it no longer did in the daily-caffeine group (Beaumont and colleagues, 2017^[23]). In 11 active people given 3 mg/kg every day for 20 days, the benefit held for about 15 to 18 days but grew smaller over time (Lara and colleagues, 2019^[24]).

The 2021 position stand calls the evidence mixed, with no clear answer either way. It points out that self-reported intake is unreliable, and that the cyclist trial asked riders to skip caffeine for 24 hours first. For heavy users, part of the boost may then have been relief from withdrawal. Still open: whether long-term heavy users would gain more after a longer break, and whether anyone's self-reported intake is accurate enough to settle this. For a daily coffee drinker deciding on a break before a race, the 2022 analysis found the boost whether people had gone without caffeine for less than a day or for more than two days before testing.

Do your genes decide whether caffeine works for you?

Genes may explain part of why some people respond and others do not, but the trials disagree. The main candidate is CYP1A2, the gene for the liver enzyme that breaks down most caffeine. In the largest trial, 101 competitive male athletes took 0, 2 or 4 mg/kg before a 10-km cycling time trial. Riders with the AA version, often called fast metabolizers, were 4.8% faster at 2 mg/kg and 6.8% faster at 4 mg/kg. Riders with the CC version, the slowest metabolizers, were only 8 of the 101, and they were 13.7% slower at 4 mg/kg. Riders with one of each (AC) saw no change (Guest and colleagues, 2018^[25]).

Several later trials found no gene effect on performance. They include 40 male cyclists at 5 mg/kg (Glaister and colleagues, 2021^[26]), 18 adults at 3 mg/kg, 31 professional handball players at 3 mg/kg, and 90 adolescent athletes at 6 mg/kg. A second gene, ADORA2A, carries the instructions for one of the adenosine receptors that caffeine latches onto in the body. It was linked to non-response in a 12-woman pilot; a later trial found 20 men carrying the supposedly less-responsive version improved on 21 of 25 measures (Grgic and colleagues, 2020^[27]). The 2021 position stand ties ADORA2A more firmly to caffeine-related anxiety and sleep problems than to performance.

Researchers on both sides of the gene question have ties to genetic-testing companies. Nanci Guest, who led both the 2018 trial that found a gene effect and the 2021 position stand, consults for and advises Nutrigenomix, a company that sells genetic tests. Craig Pickering, a co-author of the ADORA2A trial that found the supposed non-responders still responded, and of reviews on this question, is a former employee of the genetic-testing company DNAFit.

Still open: nobody has yet run a trial large enough to check the CC result, which rests on 8 riders, or to show whether genes matter more for sleep and anxiety than for performance.

Is the evidence different for women?

For endurance, women get about the same benefit as men. For lifting and sprinting, the evidence is thinner and mixed. Most of the studies pooled in the big meta-analyses were done in young men, but women-only trials and meta-analyses now exist.

In a direct comparison, 11 women and 16 men, all trained cyclists or triathletes, took 3 mg/kg. In a cycling test where everyone had to complete the same set amount of work as fast as possible, women finished 4.3% faster and men 4.6% faster (Skinner and colleagues, 2019^[28]). A review of 10 studies that compared the sexes found no difference in any of the endurance comparisons. In 4 of 7 comparisons of short, intense work (power, weight lifted, sprint speed), the benefit was larger in men (Mielgo-Ayuso and colleagues, 2019^[29]).

Women-only pooled data point toward similar small effects. A 2021 meta-analysis of 8 studies in women found small strength and muscular-endurance benefits, which the authors call similar to men's, with the upper body clearer than the lower body (Grgic and Del Coso, 2021^[30]). A 2026 meta-analysis of 20 trials in women found a small overall benefit in every menstrual-cycle phase and in hormonal-contraceptive users, with no clear difference between them. The authors rated certainty as low to very low (Zhu and colleagues, 2026^[31]). One group of 13 trained women tested at 3 mg/kg across three cycle phases also improved similarly in each.

Still open: whether sprint and lifting gains are smaller in women, and whether hormonal contraception changes the response.

Does caffeine dehydrate you during exercise?

No. Caffeine has a small effect on urine output at rest, and that effect largely disappears once you are exercising. A 2015 meta-analysis of 16 studies (where the middle dose was 300 mg) found a clear increase at rest, while the exercise effect was too small to tell apart from chance (Zhang and colleagues, 2015^[32]). Across the combined studies, women showed a larger effect on urine volume than men, but the review analyzed sex and activity separately. That comparison does not tell us whether exercising women lose more urine. The NIH exercise sheet, citing this work, says caffeine does not cause enough extra urine or sweat during exercise to harm fluid balance.

What does a late-day training dose cost in sleep?

A full sports dose late in the day can disrupt sleep even many hours later, and you may not notice it. In a 2025 trial, 23 men took either 100 or 400 mg of caffeine 12, 8 or 4 hours before bed, and their sleep was measured at home. At 100 mg, the study detected no statistically significant sleep change at any of those times, even 4 hours before bed. The 400 mg dose reduced deep sleep by about 30 minutes at 4 hours, 15 minutes at 8 hours and 21 minutes at 12 hours before bed, compared with placebo. At 4 hours it also cut total sleep by about 51 minutes; the smaller total-sleep reductions at 8 and 12 hours were uncertain. Sleep was more broken up at 4 and 8 hours, and the men rated its quality as worse only at the 4-hour mark (Gardiner and colleagues, 2025^[33]).

Most sports doses sit between those two amounts: 3 mg/kg at 70 kg is 210 mg, and 6 mg/kg is 420 mg. Training studies point the same way. In the morning-versus-afternoon squat trial described earlier (13 men, 6 mg/kg), sleep was affected in 1 man after a morning dose and in 6 after a dose at 6 p.m. In 90 athletes who were light caffeine users, 31.2% reported insomnia after 3 mg/kg in an energy drink, compared with 10.4% after the decaf version. They drank it an hour before a training session, but the study did not report the time of day, so it shows how often one moderate dose costs sleep, not how late is too late (Salinero and colleagues, 2014^[34]). In 5 people tested under controlled conditions, about 200 mg taken 3 hours before bed delayed their body clock, measured by the timing of the sleep hormone melatonin, by about 40 minutes (Burke and colleagues, 2015^[35]).

Caffeine's half-life is about 4 to 5 hours, according to the NIH exercise sheet, so half of a 6 p.m. dose is still in your body around 10 or 11 p.m. If you train in the evening, your choices are a smaller dose, an earlier one, or none on those days.

Does caffeine interfere with creatine?

Taking caffeine every day during creatine loading—the initial period of higher daily creatine doses—blunted the creatine benefit in one small early trial. Later work found that a single caffeine dose before training, taken after loading, still helps, and no review shows consistent harm.

The worry started with a 1996 trial in 9 men. They loaded creatine (0.5 g per kg a day for 6 days), with or without 5 mg/kg of caffeine every day. Creatine alone raised knee-extension strength by 10% to 23%; with daily caffeine, strength did not improve, even though muscle creatine rose in both conditions (Vandenberghe and colleagues, 1996^[36]). A 2002 trial in 10 men found the two pulling in opposite directions inside the muscle. In a test of 30 electrically triggered thigh contractions, creatine made the muscle relax faster between them, while three days of caffeine made it relax more slowly, most of all as the muscle tired, and cancelled creatine's effect. The force of each contraction did not change in any condition, so the trial could not show whether slower relaxation would cost anyone strength in the gym (Hespel and colleagues, 2002^[37]).

A dose before training works differently. In 14 trained men who had already loaded creatine without caffeine, 5 mg/kg an hour before an all-out run extended time to exhaustion to about 222 seconds, compared with 198 seconds on placebo (Doherty and colleagues, 2002^[38]). A 2023 review of 10 studies found that caffeine taken after loading added a benefit in 3 of 5 studies. Daily caffeine during loading added a benefit in 1 of 5. None of the studies showed consistent harm (Marinho and colleagues, 2023^[39]).

A 2016 trial that gave 54 men about 300 mg of caffeine a day, as coffee or as powder, during a 5-day creatine load cannot answer the question. In that trial, creatine alone did not beat placebo either, and nobody took caffeine on test day (Trexler and colleagues, 2016^[40]).

The 2021 position stand suggests taking caffeine and creatine separately or avoiding high caffeine intake while using creatine. It does not say what "separately" means, such as different times of day or different days, and the 2017 creatine position stand it cites for that advice does not mention caffeine. The stand also urges caution with daily caffeine above 9 mg/kg alongside creatine. That line sits well above the 5 mg/kg a day that interfered with creatine in the 1996 and 2002 trials, so staying under 9 mg/kg does not keep you below the dose where interference showed up. No trial has tested whether ordinary coffee-level intake matters during routine daily creatine use, without a loading phase.

What you may have heard on podcasts

Three podcast guests have made caffeine claims, on women, late-day doses and creatine. Sims and Candow go further than the trials; Israetel's warning fits them.

Stacy Sims, The Diary Of A CEO, 6 January 2025. Sims said the caffeine performance research is male, that a woman using caffeine has to eat with it, and that lifting fasted on black coffee puts a woman's lean mass at risk. The women-only trials and meta-analyses described above found endurance and strength benefits similar to men's, so "the research is all male" no longer holds. No trial has tested her eating rule, which remains her advice.

Mike Israetel, Renaissance Periodization, 10 August 2026. Israetel named caffeine as an honorable mention, outside his top five supplements. He said it reliably helps training and can cut appetite during a diet. He said it can also disturb sleep, anxiety, heart rate and blood pressure, and that taking it too late in the day trades a better session for worse recovery. His warning about late doses matches the direction of the sleep trial above.

Darren Candow, two interviews. Candow is a creatine researcher who disclosed an advisory-board role with Alzchem, which makes Creapure creatine. On FoundMyFitness (31 March 2025), he said caffeine seems to interfere with creatine "if the dose is over 250 milligrams or even higher," that one or two ordinary cups should be fine, and that he personally does not mix the two. On The Diary Of A CEO (15 June 2026), he said coffee or tea is fine as long as the caffeine stays under 350 mg. So the same guest gave two different cutoffs, 250 mg in one interview and 350 mg in the other, and neither comes from a named trial. For comparison, the 1996 trial that found interference gave 5 mg/kg every day, which is 350 to 400 mg at 70 to 80 kg.

The bottom line

Studies of caffeine for exercise use 3 to 6 mg/kg, with benefits also seen at about 3 mg/kg or less. Take it about an hour before (gum can be closer). Count every source, because a full sports drink dose or a pre-workout plus morning coffee can pass 400 mg. Expect the clearest benefit in endurance sessions and smaller gains in lifting and team sports. If you train late, weigh the session boost against how it might interfere with your sleep and recovery.

Help us improve this article

Spotted an error or found a source we should consider? Tell us which part of this article needs another look and include a link if you have one. Please leave out personal health details and lab results; we can’t answer individual medical questions.

Your name

Email address

What would you like to share?Select a topicCorrectionSource or research suggestionArticle requestAccessibility issueBusiness inquirySomething else

Your message

Continue to email →Email us directly