Caffeine and Performance
Caffeine is the second best evidenced supplement in sport. It also makes a minority of people measurably worse, which nobody selling it mentions.
Last updated: 30 September 2026
Overview
Caffeine and creatine are the two supplements with genuinely strong performance evidence, and caffeine is the one almost everyone is already taking.
This article is the detail that the supplement overview only had room to summarise: what the dose is, what it does and does not do, why a minority of people get worse on it, and what the safety numbers actually are.
Quick Answer
Caffeine improves exercise performance at doses of 3 to 6mg per kg of body mass, taken around 60 minutes before exercise. Benefits are largest for endurance, around 2 to 4 percent, and smaller for strength. Doses of 9mg per kg bring side effects without extra benefit.
Key Points
- 3 to 6mg per kg, about 60 minutes before. Minimum useful dose may be as low as 2mg per kg.
- Endurance gains 2 to 4 percent; strength gains are small.
- 9mg per kg gives side effects and no more performance.
- A minority of people are slower on caffeine, for genetic reasons.
- Sleep costs will outweigh any of this. Mind the cut-off.
Dose and timing
| Effective dose | 3 to 6mg per kg body mass |
| Possible minimum | As low as 2mg per kg |
| Too much | 9mg per kg: side effects, no added benefit |
| Timing | About 60 minutes before, for peak blood levels |
| Faster option | Chewing gum absorbs through the mouth, so needs less lead time |
For a 70kg person, 3 to 6mg per kg is 210 to 420mg. For scale, NHS figures put a cup of instant coffee at about 100mg, with filter and coffee-shop coffee often higher, a cup of tea at 75mg, a can of cola at 40mg and a 250ml energy drink at 80mg.
Start at the bottom of the range. There is no prize for the high end and the side effects are real.
What it does, by exercise type
| Type | Benefit |
|---|---|
| Endurance | Most consistent, moderate to large. Around 2 to 4 percent |
| Muscular endurance | Modest, around 6 to 7 percent, inconsistent between studies |
| Strength | Small. Roughly 2 to 7 percent, small effect sizes |
| Sprinting and power | Mixed. Repeated sprint results are inconsistent |
| Attention and vigilance | Ergogenic in most people |
The pattern is worth reading properly: caffeine is an endurance supplement that also helps a bit with lifting, not a strength supplement. Much of its value in the gym is the perceived-effort effect — the session feels easier at the same load, which is why it works best on days you do not want to be there.
The part nobody selling it mentions
Response to caffeine is partly genetic, and the differences are not small.
For the CYP1A2 gene, which governs how fast you clear caffeine, one study at 4mg per kg found:
| Genotype | Result |
|---|---|
| AA, fast metabolisers | 6.8% improvement |
| AC | No effect |
| CC, slow metabolisers | 13.7% impairment |
A slow metaboliser taking a standard pre-workout is not getting a smaller benefit. They are getting measurably worse, at the dose everybody recommends. The effect is most pronounced in longer efforts where fatigue accumulates.
There is also weaker evidence around ADORA2A, which appears to modulate how much caffeine disturbs sleep and raises anxiety.
You almost certainly have not been genotyped, so the practical version is: if caffeine before training reliably makes you jittery, anxious and worse rather than sharper, believe yourself. That is data, and it beats the average from a meta-analysis.
Tolerance
Less settled than the confident version you have heard. The position stand describes the findings as equivocal: one study found tolerance developing after four weeks in low-habitual users, another found the effect maintained over 20 days with the magnitude fading, and more recent work found caffeine improved performance regardless of habitual intake.
Sensible reading: habitual use probably blunts the acute effect somewhat, the size of that is unclear, and there is no strong case for elaborate cycling protocols. Using less on ordinary days so it still does something on hard ones is reasonable, not obligatory.
Sleep, which is the real trade
Caffeine lengthens the time it takes to fall asleep, reduces deep non-REM sleep, and lowers sleep efficiency. Its half-life means an afternoon dose is still working at bedtime.
No 3 percent improvement in a set of squats is worth an hour of lost sleep, because sleep is doing more for your training than the caffeine is. If you train in the evening, this is the argument for training without it.
A cut-off of at least 8 hours before bed is the usual advice, and further if you are sensitive.
Safety
| Population | Guidance |
|---|---|
| Adults | Up to 400mg a day does not raise safety concerns |
| Adults, single dose | Up to 200mg at once does not raise safety concerns |
| Pregnancy | Up to 200mg a day |
| Under 18 | Lower thresholds. Energy drinks are not a training supplement |
Notice the tension: a 5mg per kg dose for an 80kg lifter is 400mg in one go, which is double the 200mg single-dose figure and the whole daily allowance at once. The performance literature and the food-safety guidance are answering different questions, and the sensible move is the lower end of the performance range rather than the top.
If you are pregnant, the 200mg daily figure is the one to use, and exercise in pregnancy covers the rest.
Forms
| Form | Notes |
|---|---|
| Anhydrous capsules | Precise dose, cheapest, peak at 30 to 120 minutes |
| Coffee | Works, dose varies a lot by bean and brew |
| Chewing gum | Absorbed through the mouth, faster, less lead time needed |
| Gels and bars | Evidence growing, mainly for endurance |
| Pre-workout blends | Caffeine plus filler, at a markup |
Common Mistakes
| Mistake | Why it causes problems | Better approach |
|---|---|---|
| Going straight to a high dose | More side effects, no more benefit | Start at 3mg per kg |
| Taking it 5 minutes before | Peak is around 60 minutes | Time it properly |
| Evening training on a big dose | Costs sleep, which costs more | Train without it, or dose early |
| Assuming you are a responder | A minority get measurably worse | Test it against caffeine-free sessions |
| Paying for a pre-workout blend | Mostly caffeine and filler | Buy caffeine |
| Expecting a strength supplement | Effects on strength are small | Use it for endurance and hard days |
Frequently Asked Questions
How much caffeine should I take before training?
3 to 6mg per kg of body mass, about 60 minutes beforehand. Start at the low end and note that the food-safety single-dose figure is 200mg.
Does caffeine help lifting?
A little. Strength improvements are small, around 2 to 7 percent with small effect sizes. Its clearest benefits are in endurance and in how hard the session feels.
Do I need to cycle off caffeine?
The evidence on tolerance is equivocal and there is no strong case for a cycling protocol. Keeping ordinary days lower so hard days still benefit is reasonable.
Can caffeine make performance worse?
Yes, for some people. In one study, slow-metabolising CC genotypes were 13.7 percent worse at 4mg per kg. If it consistently makes you feel bad and perform badly, that is real.
How much caffeine a day is safe?
Up to 400mg a day for adults, and up to 200mg in a single dose, do not raise safety concerns. In pregnancy the figure is 200mg a day.
Sources
Guest et al. (2021), International Society of Sports Nutrition position stand: caffeine and exercise performance - https://pmc.ncbi.nlm.nih.gov/articles/PMC7777221/
EFSA on the safety of caffeine - https://www.efsa.europa.eu/en/topics/topic/caffeine
NHS advice on foods to avoid in pregnancy, including caffeine - https://www.nhs.uk/pregnancy/keeping-well/foods-to-avoid/