Ask anyone why they drink eight glasses of water a day and you get a shrug, a guess, or a confident nod towards “science”. Nobody has ever been able to name the actual study.

There isn’t one. That is the short version of what turned up when Heinz Valtin, a kidney physiologist at Dartmouth Medical School who had spent a career studying how the body manages water, went looking for the evidence behind “drink eight glasses of water a day”. He searched the electronic databases and dug through older literature that never made it online, then rang nutritionists who specialised in thirst and drinking. His 2002 review in the American Journal of Physiology reported, flatly, that no scientific studies supported the advice.

The sentence that got cut in half

Best guess at an origin: a 1945 recommendation from the US Food and Nutrition Board, which put a suitable daily allowance for adults at about 2.5 litres. Two and a half litres is, near enough, eight glasses. The very next sentence pointed out that most of that quantity is already contained in prepared foods.

That clause did not survive the retelling.

Aaron Carroll, a physician who writes on medical myths, traced the same 1945 wording for the New York Times in 2015 and reached the identical conclusion. Readers took the number, skipped the qualifier, and an estimate of total water intake hardened into a drinking quota. Eight decades of repetition later, the qualifier is still missing and the number is still being quoted.

What the official guidance actually says

Anyone hoping for one clean number will be let down. When the Institute of Medicine of the US National Academies published intake levels for water in 2004, its panel concluded that the vast majority of healthy people meet their hydration needs by letting thirst be the guide. Its headline figures, roughly 3.7 litres a day for men and 2.7 for women, count every beverage and every mouthful eaten, not just glasses of plain water, and roughly a fifth of that arrives through food alone. Caffeinated drinks count towards it. Valtin’s review argued that beer, in moderation, does too.

How much people get through in practice

Do those averages hide a lot of variation? Comfortably. A 2022 study in Science measured water turnover in 5,604 people, aged eight days to 96 years, across 23 countries, using deuterium tracking. Daily turnover ranged from about one litre to more than ten, moving with body size, physical activity, air temperature, humidity, altitude, pregnancy and even national development level. Its authors wrote that one size clearly does not fit all, and that the eight-glass suggestion has no objective evidence behind it. Other researchers wrote in to the journal to argue that turnover and optimal intake are two different quantities, which is fair. One paper, however large, is still one paper.

The claims that come attached to the rule

Flushing toxins. Better skin, fewer headaches, easier weight loss. Dan Negoianu and Stanley Goldfarb of the University of Pennsylvania went through the published clinical evidence for each of these in a 2008 editorial for the Journal of the American Society of Nephrology and came up empty on all four. They were careful with the distinction, though: nobody had demonstrated a lack of benefit either, and mostly there was just very little research to draw on.

At the far end of the scale, the research exists and it is grim. Christopher Almond and colleagues at Children’s Hospital Boston tested blood from 488 runners after the 2002 Boston Marathon and reported in the New England Journal of Medicine that 13 per cent finished with hyponatraemia, meaning blood sodium diluted to a dangerous level. The strongest predictor was weight gained during the race.

Putting on weight over 42 kilometres means drinking faster than you sweat.

Why a rule with no parents refuses to die

Eight is a wonderfully commercial number. It fits on a label, divides neatly into a working day, sells bottles and apps and hourly reminders, and offers the small daily satisfaction of a box ticked. Thirst offers none of that. It asks for no equipment, produces no streak to maintain, and gives a person nothing to be virtuous about in front of colleagues.

Which may be the real reason a 1945 clause about prepared foods never stood a chance.