The advice sounds responsible, which is most of the reason it survives. One glass of water for every cup of coffee, because the coffee does not count. A great many people have organised their desks around that sentence. It rests on a real effect that has genuinely been measured, and then stretched a long way past what the measurement supports.
Coffee counts toward your daily fluid intake. In a randomised cross-over trial in habitual coffee drinkers, three days of moderate coffee produced no measurable difference in hydration status compared with the same volume of water. Caffeine does have a mild diuretic action. In people who have abstained for several days and then take a large single dose, it is measurable in urine output. In people who drink coffee regularly, a review of the fluid-balance literature found that response is much diminished, and at the size of a normal serving it does not show up at all.
That is the answer. The rest of this is why it is the answer, and the places where it stops being true.
What caffeine does to the kidney
Caffeine blocks adenosine receptors. That is the same action that keeps you awake, and the kidney carries those receptors too. Maughan and Griffin's review of the caffeine and fluid-balance literature sets out what follows: blocking those receptors lowers the fraction of sodium reabsorbed in the proximal tubule and the distal nephron, and water leaves with the sodium. The result is a short window of increased urine production. The mechanism is not in dispute and nobody in this argument denies it exists.
Two things blunt it. The first is dose. That review, published in the Journal of Human Nutrition and Dietetics in 2003, is a narrative review of the caffeine and fluid-balance literature up to that point. It found a short-term rise in urine output was reliably produced by large acute doses, at least 250 to 300 mg, in people who had been off caffeine for days or weeks beforehand. At doses equivalent to a standard serving of coffee, tea or a soft drink, it found no diuretic action at all. The USDA reference value for an 8 fl oz cup of brewed coffee is about 96 mg. Your cup is not the reference cup, which is the subject of the last section of this article, but the gap between one mug and 250 mg is wide.
The second is tolerance, and it is the part the folk rule ignores completely. The same review found the tolerance that develops in regular tea and coffee drinkers is substantial, which is why the effect is produced most reliably in people deprived of caffeine first. Most of the older experiments that produced dramatic-looking diuresis had asked participants to abstain entirely for days or weeks, which is a sensible way to isolate a drug effect and a poor model of anybody's actual Tuesday.
So the honest version of the mechanism is this. Caffeine is a diuretic in the pharmacological sense. Coffee, drunk the way coffee drinkers drink it, is not a net dehydrating beverage, because the drug effect is small at these doses, it habituates, and the drink itself is mostly water.
The trial that tested it directly
Sophie Killer, Andrew Blannin and Asker Jeukendrup ran the experiment the question deserves, published in PLOS ONE in January 2014. It was a randomised counterbalanced cross-over trial: fifty male habitual coffee drinkers each completed both arms, three days on four 200 ml mugs of coffee a day and three days on the same volume of water, ten days apart, with the order of the arms swapped between participants. The coffee arm delivered about 4 mg of caffeine per kilogram of body weight a day, which is a real dose. It is simply not a single dose.
The design is what makes it useful. Because every participant did both conditions, each man is his own control, which removes the between-person variation that swamps small effects in hydration research. Total body water was measured by isotope dilution rather than inferred from a scale. They also took 24-hour urine volume, blood and urine osmolality, haematocrit and serum sodium.
Across every one of those measures, the coffee days and the water days were not significantly different. The paper's conclusion, stated plainly, was that there was no evidence of dehydration at moderate daily coffee intake, and that coffee provides fluid comparably to water.
Read the limits with the result. All fifty participants were men. All were already habitual drinkers of three to six cups a day, which is exactly the population in which tolerance is fully developed. Each arm ran three days, not three months. The intake was moderate by design. This trial says something precise about habituated moderate drinkers and does not license a claim about someone drinking eight double espressos after a fortnight off.
Ranking coffee against other drinks
The second useful piece of evidence comes at it from a different angle. Maughan and colleagues published a randomised trial in the American Journal of Clinical Nutrition in 2016 covering thirteen common drinks. Seventy-two fasted, euhydrated men each drank one litre of still water and one litre of three of the other beverages, on separate occasions, over thirty minutes, with urine collected for the four hours that followed. From that they built a beverage hydration index, scoring each drink against still water at the two-hour mark.
Three drinks retained more fluid than water did: oral rehydration solution, full-fat milk and skimmed milk. Coffee did not differ significantly from water. Neither did tea, cola, lager, orange juice or a sports drink. The drink people are told to offset with water performed, in the trial built to rank them, like the water they were told to offset it with.
What this does not do
None of this makes coffee a hydration strategy, and it should not be read as one. The finding is an absence: coffee does not put you in deficit. It does not put you in credit either, it carries no advantage over water, and drinking more of it to hydrate better is a misreading of the same evidence in the opposite direction. It says nothing about sleep, about heart rhythm, about anxiety, or about anyone taking medication that interacts with caffeine. It does not extend to caffeine pills or to energy drinks, where the dose per serving can be several times a cup of filter and the trials above did not test them. And it is not advice about you. If a clinician has told you to limit caffeine or to track fluids, that instruction was made with information about you that no trial of fifty habitual drinkers contains.
Where the rule came from
The eight-glasses rule that generates the offset habit has a traceable and slightly embarrassing history. Heinz Valtin, a kidney physiologist at Dartmouth, went looking for its evidence base and published the search in the American Journal of Physiology in 2002. He could not find a scientific origin for the number. The closest candidate was a 1945 US Food and Nutrition Board recommendation of roughly 2.5 litres of water a day, which went on in its very next line to point out that food already supplies most of it. The qualifier dropped off. The number travelled.
Official bodies have not repeated the error. The European Food Safety Authority's 2010 opinion on dietary reference values for water sets adequate intakes of about 2.0 litres a day for adult women and 2.5 litres for adult men, and it counts total water from all sources: drinking water, other beverages, and the moisture in food. Coffee is one of those other beverages. There is no asterisk beside it.
The dose you cannot actually know
One honest complication, and it is the reason this article will not tell you how many cups is your limit. You do not know how much caffeine is in your cup, and neither do we.
Thomas Crozier and colleagues bought espressos from twenty different commercial outlets in one city and measured them, publishing the results in Food & Function in 2012. The range ran from 51 mg to 322 mg of caffeine per serving. Same drink name, same city, six-fold spread. The drivers were the dose of ground coffee, the arabica-to-robusta ratio, and the roast and grind.
That variability is worth understanding on its own terms, because it also explains why so much shelf advice is useless. The word strong is doing four unrelated jobs at once, only one of which is caffeine. A dark roast is not carrying more caffeine than a light one by weight, whatever the taste suggests. The number printed on a supermarket bag is a roast-and-blend index set by the roaster and checked by nobody. And robusta carries roughly twice the caffeine of arabica, which is a real difference the label rarely quantifies.
For reference rather than as a personal recommendation: the European Food Safety Authority's 2015 scientific opinion on caffeine gives 400 mg a day from all sources combined for healthy adults, and 200 mg as a single dose. Those are population-level figures from a regulator, not a number for your body, and a single 322 mg espresso would exceed the single-dose figure on its own.
The practical position is unglamorous. Drink coffee because you want coffee. Drink water because you are thirsty. Stop counting one against the other, because the counting was never built on anything, and the trial that checked found nothing to count.
Health sources on this page last checked 13 August 2026. Figures from EFSA, and from named peer-reviewed trials with their designs stated. Nothing here is medical advice.