Caffeine
Best-in-class: Caffeine
The most-used and best-evidenced ergogenic aid on earth, listed on its own rather than buried inside a pre-workout blend.
Caffeine quick facts
| Suggested dose | 3–6 mg/kg, 30–60 minutes before. More is not better and the curve turns down. |
| How often | Per session, and tolerance builds with daily use |
| Who it's for | Almost any performance context. |
| Best-in-class brand | Caffeine |
Reliably works, and the main skill is not wrecking your sleep with it. The half-life is around five to six hours, so an afternoon coffee still has a quarter of its dose on board at bedtime — and it fragments deep sleep even when you fall asleep fine. Clearance is genetic and varies severalfold, which is why some people tolerate evening coffee and others genuinely can't. Delay the first cup an hour or two after waking to avoid stacking it on top of your own cortisol peak, and take periodic breaks to reset tolerance.
How Caffeine actually works
Caffeine does not create energy — it blocks the perception of not having any. It's a structural analog of adenosine and occupies A1 and A2A receptors without activating them. Adenosine accumulates through the waking day and drives sleep pressure, so blocking it removes the brake, disinhibiting dopamine and noradrenaline signaling. The debt is real: adenosine keeps accumulating behind the blockade. Tolerance develops partly by receptor upregulation, and the ergogenic effect is separate from the alertness one, involving reduced perceived exertion and direct effects on calcium handling in muscle.
Where to get Caffeine
Find Caffeine on iHerb →The evidence for Caffeine
Graded by what exists behind each claim.
✅ Clinically validated
- Overwhelming meta-analytic support for improved endurance, strength, power and reaction time.
- Reduces perceived exertion, which is likely a large part of the mechanism.
- Effect size varies with CYP1A2 genotype and with habitual intake.
📊 Correlative data
- The best observational evidence of anything in this file. Large cohorts consistently associate habitual coffee intake with lower all-cause mortality, with a broad plateau across a wide intake range — and the association holds for decaffeinated coffee too, which suggests it is not purely the caffeine. Genetic variation in CYP1A2 also shows up in cohort data as different cardiovascular associations between fast and slow metabolizers.
🧪 Theoretical / extrapolated benefits
- Adenosine receptor antagonism is the primary mechanism; the old 'fat oxidation' explanation is largely obsolete.
How to read these tiers: they say how much human evidence exists, not how well something works — and ✗ flags harm, never a disappointing trial. How the evidence tiers work →
What Caffeine actually does
Caffeine does not stimulate anything. It removes a brake. It is a competitive antagonist at adenosine A1 and A2A receptors at concentrations achieved by a normal cup of coffee. Adenosine accumulates through the waking day as a byproduct of ATP use and signals sleep pressure; blocking its receptors does not add energy, it hides the accumulating signal that energy is being spent. Everything downstream follows from that one displacement.
The two receptors do different jobs and explain the two halves of the effect. A1 antagonism in cortex and basal forebrain produces the arousal and the reduced perception of effort. A2A antagonism in the striatum disinhibits dopaminergic signaling through the A2A-D2 receptor heteromer, which is the reinforcing half and the reason caffeine is a habit rather than a tool.
The mechanisms in the textbook are the ones that do not operate. Phosphodiesterase inhibition and ryanodine receptor sensitization both require concentrations roughly twenty times higher than a dietary dose achieves, and lethal doses are reached long before them. Any explanation of caffeine's ergogenic effect that starts with cyclic AMP is describing a test tube.
The clearance enzyme is a single cytochrome and it is unusually variable. CYP1A2 performs about 95 percent of caffeine demethylation to paraxanthine, theobromine and theophylline. Its activity varies severalfold with the common -163C>A variant, is induced by tobacco smoke and cruciferous vegetables, and is inhibited by oral contraceptives, fluvoxamine and ciprofloxacin. That is why plasma half-life spans 2.5 to 10 hours in healthy adults and reaches 15 hours in late pregnancy.
Adaptation is receptor-level and it is why the baseline matters. Chronic exposure upregulates adenosine receptor density, so a habitual user needs caffeine to return to the alertness a non-user has for free. That is not tolerance to a benefit; it is the creation of a deficit that the next dose repairs.
Cell, rodent, human — and where it stops
Cell to human is short. The problem on this page is not species or dose; it is that the comparison group in most trials was in withdrawal.
The ergogenic effect in acute trials is one of the best established in sports science. The position stand summarizes doses of 3 to 6 mg per kg improving muscular endurance, movement velocity, sprint and endurance performance, with the effect present in trained and untrained people and at doses that are ordinary rather than extreme Guest 2021.
The mood and cognition literature is where the surrogate breaks. When habitual consumers are deprived overnight and then tested, caffeine improves alertness and performance — against their withdrawn state. Testing genuine non-consumers removes most of it: caffeine produced no reinforcing, mood or psychomotor performance benefit in habitual non-consumers Rogers 2003, and the size of the apparent benefit tracked with the level of abstinence in habitual users Yeomans 2002.
That is the cleanest example of a moving baseline in this catalog. The number that moved — a reaction time, an alertness rating — moved relative to a state the drug itself created. A morning coffee restoring a person to where a non-drinker starts is a real subjective experience and it is not a performance-enhancing effect.
The genetic marker was supposed to explain the non-responders and it does not, cleanly. A systematic review with meta-analysis found the evidence that CYP1A2 genotype determines the ergogenic response inconsistent Wang 2023, and a more recent controlled study examined the genotype question again in trained individuals for strength and endurance endpoints Montalvo-Alonso 2026. Genotype is a marker with a mechanism and, so far, without a reliable prediction.
Where the outcome transfers with no ambiguity at all is sleep. 400 mg taken 6 hours before bed reduced total sleep time by more than an hour by objective polysomnography, and participants did not notice Drake 2013. The subjective read-out and the measured one disagreed, in the direction that lets a habit persist.
Caffeine — which form, and does it matter
Anhydrous caffeine and coffee are not interchangeable exposures. Anhydrous powder is a defined dose with a rapid, reproducible rise. Coffee delivers a variable dose — roughly 70 to 140 mg per cup depending on bean, roast and method — alongside chlorogenic acids that slow gastric emptying and blunt the peak. A trial using anhydrous caffeine is not a trial about coffee, and the trials mostly used anhydrous.
The pharmacokinetics are simple and that is why the timing advice is firm. Oral caffeine is absorbed essentially completely, peaks at 30 to 60 minutes, and shows negligible first-pass metabolism — oral bioavailability is close to 100 percent. Clearance is almost entirely hepatic through CYP1A2, with under 3 percent appearing unchanged in urine, so renal clearance is not a meaningful route Guest 2021.
The half-life is the number that decides the evening. At a 5-hour half-life, 200 mg at 2 pm leaves 100 mg at 7 pm and 50 mg at midnight. The polysomnography study is the empirical version of that arithmetic: a dose 6 hours before bed still cost over an hour of sleep Drake 2013. Anybody with a slow CYP1A2 phenotype is working from a longer clock than that.
The primary metabolite is now a product in its own right. Paraxanthine accounts for roughly 80 percent of caffeine's demethylation and carries much of the adenosine antagonism with a different peripheral profile; a randomized crossover has compared the two directly on rowing performance and on sleep quality Bingol Diedhiou 2026. A person taking paraxanthine is taking a metabolite with its own clock rather than a cleaner caffeine.
Delivery format changes the peak and therefore the experience. Chewing gum delivers buccally and peaks faster than a capsule; extended-release formulations flatten the curve; energy drinks combine caffeine with sugar and with taurine and B vitamins whose contribution is unquantified. A label that gives milligrams without a release profile has specified the dose and not the exposure.
What would have to be true, and how you would know it was not
1. Predict the ergogenic effect survives in a rested, habituated person and the cognitive one mostly does not. At 3 to 6 mg/kg taken 60 minutes before, predict a measurable improvement in time to exhaustion or repeated sprint performance Guest 2021, and predict that a well-slept habitual user gets far less subjective lift than the same person tested after overnight abstinence Rogers 2003.
2. The self-experiment that separates benefit from withdrawal relief. Predict that a two-week complete washout, followed by a blinded test of 200 mg against placebo on separate days, shrinks the measured effect substantially compared with the same test done on day one of abstinence Yeomans 2002. That is falsifiable at home and almost nobody runs it.
3. The prediction that cuts against the product. Predict that actigraphy or a polysomnograph shows reduced total sleep time from a dose taken 6 hours before bed even when the person reports sleeping normally Drake 2013. A study showing no objective sleep cost at that interval would falsify this page's central caution.
4. Predict genotype does not tell you what to do. Predict that a CYP1A2 result does not reliably predict the size of the ergogenic response, because the meta-analytic evidence is inconsistent Wang 2023 Montalvo-Alonso 2026. A large, pre-registered, genotype-stratified trial with a clear interaction would overturn that and would be worth running.
5. Predict the withdrawal, and use it to size the habit. Predict headache, fatigue and low mood beginning 12 to 24 hours after the last dose, peaking at 20 to 51 hours and resolving over 2 to 9 days. Predict severity scales with habitual intake. The presence and size of that syndrome is the most honest measure of how much of the daily benefit was repair.
What nobody has tested yet
Nobody has run the trial that would settle the cognition question. A properly powered study in lifelong non-consumers, with a genuine dose-response and objective endpoints, would establish whether caffeine has any net cognitive benefit at all or only restores a self-created deficit Rogers 2003. Two decades after the question was framed clearly, it is unanswered.
The paraxanthine claim is one crossover old. Whether the metabolite genuinely delivers ergogenic effect with less sleep cost, as its marketing claims, rests on a very small evidence base Bingol Diedhiou 2026. Replication with polysomnography rather than sleep questionnaires is the obvious next step.
The genotype question needs a different design. Existing trials are underpowered for a gene-by-treatment interaction, which needs several times the sample size of a main-effect trial Wang 2023. Until somebody funds that, the commercial genotype tests sold for caffeine response are ahead of the evidence.
And nobody knows the long-term consequence of receptor upregulation. Adenosine receptor density rises with chronic exposure and no human study has characterized how long it takes to return to baseline after cessation, or whether it ever fully does in somebody with decades of use.
Caffeine — its own safety story, not its category's
The dose-limiting effects arrive long before anything dangerous. Anxiety, tremor, palpitations and gastrointestinal upset are dose-related and are how most people find their ceiling. 400 mg a day is the commonly cited limit for healthy adults and 200 mg in pregnancy; single doses above roughly 9 mg/kg add side effects without adding performance Guest 2021.
Anhydrous powder is the genuine hazard in this category. A teaspoon of pure caffeine powder is roughly 5 grams — more than twenty cups of coffee — and deaths have occurred from measuring bulk powder with a kitchen spoon. This is one of the few supplements where the format, not the molecule, has killed people.
The sleep cost is the effect most users do not attribute correctly. Objective sleep loss occurs at intervals people consider safe and without a matching subjective complaint Drake 2013. A person treating morning fatigue with more caffeine is usually treating the previous day's dose.
Interactions run through one enzyme and are therefore predictable. Fluvoxamine, ciprofloxacin, oral contraceptives and cimetidine inhibit CYP1A2 and can double or triple caffeine exposure at an unchanged intake. Smoking cessation removes a potent inducer, so caffeine concentrations rise sharply within days of quitting, which is a frequently missed cause of jitteriness and insomnia in that setting. Theophylline shares the pathway. Adenosine is used diagnostically in cardiac stress testing and caffeine directly antagonizes it, which is why abstinence is required before the test.
Who should be careful. Anyone with an arrhythmia, uncontrolled hypertension, an anxiety disorder or a sleep disorder; anyone pregnant, where clearance slows to around 15 hours by the third trimester; and adolescents, in whom energy drink intake is the commonest route to a large unintended dose. Nothing here is medical advice or diagnosis, and these statements have not been evaluated by the Food and Drug Administration.
Sources read for this page
- Guest NS. International society of sports nutrition position stand: caffeine and exercise performance. J Int Soc Sports Nutr 2021 · PMID 33388079
- Wang J. Does ergogenic effect of caffeine supplementation depend on CYP1A2 genotypes? A systematic review with meta-analysis. J Sport Health Sci 2023 · PMID 38158179
- Montalvo-Alonso JJ. CYP1A2 Genotype and the Ergogenic Effect of Acute Caffeine Intake on Muscular Strength and Endurance Performance in Trained Individuals. Scand J Med Sci Sports 2026 · PMID 41627185
- Rogers PJ, et al. Absence of reinforcing, mood and psychomotor performance effects of caffeine in habitual non-consumers of caffeine. Psychopharmacology 2003 · PMID 12601503
- Yeomans MR, et al. Effects of caffeine on performance and mood depend on the level of caffeine abstinence. Psychopharmacology 2002 · PMID 12424547
- Drake C. Caffeine effects on sleep taken 0, 3, or 6 hours before going to bed. J Clin Sleep Med 2013 · PMID 24235903
- Bingol Diedhiou A. Comparative effects of caffeine and paraxanthine on rowing performance and sleep quality: a randomized crossover study. J Int Soc Sports Nutr 2026 · PMID 41918248
How you would know if it worked
There is no blood test for this one. That is not a criticism — it is a fact about the effect, and it changes how you should judge it.
- What to watch: One performance number and one sleep number, because this moves both and only one of them is why you bought it. Take a repeatable performance measure and record it; log the time you fell asleep and the times you woke. The effect on effort is real and large enough to see in a single session. The cost arrives eight hours later, and anyone who tracks only the first half of that trade concludes it is free.
- How long before it means anything: One session for the ergogenic effect, at 3 to 6 mg per kilogram 30 to 60 minutes before. Three weeks for the honest version, because tolerance builds with daily use, and the picture at week three is the one you will actually be living with.
- What will fool you: Withdrawal wearing the costume of a benefit. If you drink coffee daily, a dose taken after a gap is reversing a deficit rather than adding anything — which is why habitual users overestimate what caffeine does, and why the fairest test starts from several caffeine-free days. Then timing: the half-life is around five hours, so an afternoon dose measurably degrades that night's sleep even in people certain that it does not, and sleep is the strongest recovery intervention there is. Weigh powder on an accurate scale or use capsules — anhydrous caffeine has killed people through kitchen-spoon measuring errors.
Run it one variable at a time. Starting three things in one week means a result you cannot attribute, which is the same as no result.
Caffeine — safety & side effects
- Jitteriness, anxiety, raised heart rate, GI upset and disrupted sleep — all dose-related, all familiar. Caffeine's half-life is 5–6 hours, so an afternoon dose is still working at bedtime even if you fall asleep fine.
- Dependence and a real withdrawal syndrome — headache, fatigue, low mood for several days. Taper rather than stopping abruptly.
- Powdered caffeine has killed people. A teaspoon can be a lethal dose; never measure bulk powder by volume.
- Interacts with ciprofloxacin, fluvoxamine and theophylline (raising levels), and additively with other stimulants. Caution in arrhythmia, anxiety disorders and pregnancy (limit 200 mg/day).
Not medical advice. If you take prescription medication or have a diagnosed condition, check this against it with a pharmacist or doctor — pharmacists are underused and free.
Build your foundation with Coach Cam
The full Supplement Vault — 371 products across 14 categories with clinical, correlative & theoretical evidence, plus my Thorne partner links — lives inside Skool alongside 278 peptides.
Join Skool — $10/mo →Bloodwork to run alongside Caffeine
Baseline first, then again at 8–12 weeks.
| Marker | What it’s watching for |
|---|---|
| Complete Blood Count (CBC) with Differential | Hemoglobin sets your ceiling for endurance work |
| Ferritin | Low iron limits performance long before anemia shows |
| Total Testosterone | Drops under genuine overreaching |
| Comprehensive Metabolic Panel (CMP) | Kidney and electrolytes under heavy training load |
The Athletic Performance & Recovery panel covers these in one order — 12 markers, $207.90 with the discount applied.
Check results you already have → · All 103 markers A–Z
Caffeine — frequently asked questions
What is Caffeine?
The most-used and best-evidenced ergogenic aid on earth, listed on its own rather than buried inside a pre-workout blend.
What is the suggested dose of Caffeine?
3–6 mg/kg, 30–60 minutes before. More is not better and the curve turns down. This is a general reference for education only — statements have not been evaluated by the FDA and this is not medical advice.
What are the researched benefits of Caffeine?
Overwhelming meta-analytic support for improved endurance, strength, power and reaction time.
Who is Caffeine for?
Almost any performance context.
Where can I buy Caffeine?
Coach Cam sources Caffeine from vetted, top-rated brands on iHerb — use the buy link on this page.
Caffeine inside a finished plan
One arm of 1 Protocol Blueprint, free to read in full.
What Caffeine is used for
Caffeine appears under 3 goals in the goal router.
Related Performance supplements
Where this goes next
Caffeine is the buffering arm of this plan. The page above is the free breakdown of one compound; the plan it belongs to — the dosing, the order to correct things in, the week-by-week schedule and what to retest — is a lesson inside Skool.