Caffeine, and the difference between lasting longer and going faster

Twelve athletes took caffeine before riding a bike until they could not keep going. It made no difference to how long they lasted. DGrade D, Preliminary. Human evidence with major recorded weaknesses, or non-human, uncontrolled or unverified evidence.PMID:42196972

That is the whole of it, and it is worth a page because of what sits beside it in the graph. The same compound, tested a different way, comes out of a much larger body of evidence looking useful. Both records are honest. They are answers to different questions, and the space between them is most of what there is to understand about caffeine and exercise.

What the trial did

The participants were eumenorrheic — they had regular menstrual cycles — and each was tested three times, at a different point in her cycle, because the researchers wanted to know whether the phase of the cycle changed the answer. The phase made no difference either. DGrade D, Preliminary. Human evidence with major recorded weaknesses, or non-human, uncontrolled or unverified evidence.PMID:42196972

Each athlete took 400 mg of caffeine an hour before the test. That is roughly four mugs of brewed coffee, taken at once, which is a large single dose rather than a habitual one.

“Time to exhaustion was not significantly influenced by CAF supplementation or menstrual cycle phase (p > 0.05).”

Twelve people is a small number, and the mark beside those sentences says so. A single trial of twelve cannot settle a question. What it can do is report what happened in those twelve, and what it reports is a null — a result in which nothing measurable occurred.

The record that points the other way, and does not contradict it

Pooling many trials of healthy adults aged 18 to 59, caffeine taken before exercise shortened the time it took to complete a fixed-distance effort. BGrade B, Likely. Human evidence with one recorded weakness, or capped at B by its design or by unread numbers.PMID:42356375 That record carries a stronger mark than the one above, because it pools many trials rather than reporting one.

“Both low and moderate caffeine doses significantly reduced time-trial completion time relative to placebo.”

Set the two side by side and they look incompatible. Caffeine did not help these athletes keep going. DGrade D, Preliminary. Human evidence with major recorded weaknesses, or non-human, uncontrolled or unverified evidence.PMID:42196972 Caffeine did help those adults finish sooner. BGrade B, Likely. Human evidence with one recorded weakness, or capped at B by its design or by unread numbers.PMID:42356375 Both marks are on the page and neither cancels the other, because the two tests measure different things:

• Time to exhaustion fixes the workload and asks how long you can hold it. You ride at a set intensity until you stop. The number is minutes survived.
• A time trial fixes the distance and asks how quickly you cover it. You choose your own effort throughout. The number is minutes taken.

One asks you to endure at somebody else's pace. The other asks you to judge your own. A substance can act on one and not the other, and reading a null in the first as a refutation of the second treats two questions as one.

The populations differ too. The first record is trained female athletes; the second is healthy adults of both sexes across four decades of age. Neither is a statement about the other's people.

It also did not change how the effort felt

There is a familiar account of caffeine in which it works by making a given effort feel easier. That account has a record here, and the record does not support it.

Across nine studies of adult female team-sport athletes, caffeine had no effect on how hard the exercise felt, measured on a standard perceived-exertion scale. BGrade B, Likely. Human evidence with one recorded weakness, or capped at B by its design or by unread numbers.PMID:34684665

“As shown in C, nine studies evaluated RPE after performing an exercise protocol after caffeine supplementation, finding no effect of this supplement on RPE (SMD: 0.258, 95% CI: −0.048, 0.565; I2 0.26% p = 0.260).”

That one pools nine studies, which is why its mark is stronger than the twelve-athlete record at the top. The distinction matters: a null drawn from many trials says the effect is probably not there, while a null drawn from twelve people says only that this study did not find it. They are different statements and this platform draws them differently on purpose.

What none of this shows

It does not show that caffeine does nothing before exercise — one of the three records above is not a null, and it is the largest of them. It does not settle endurance in general; it reports one small trial in one group of people. And it says nothing whatever about doses taken daily over months, about untrained people, or about anyone who has not been studied.

That last case is the ordinary one. Most questions anybody actually has about a supplement have never been put to a trial, and a site that publishes only the questions with answers will read as far more certain than the evidence beneath it.

Every sentence on this page that states a finding is followed by the mark of the record it reads, and each of those links to the paper. A sentence here with no mark is not a finding — it is context, or it is the writing explaining itself.