How Much Water a Day? The 1945 Advice’s Last Sentence Was About Thirst.

In 2002 a physiologist called Heinz Valtin went looking for the study behind “drink eight glasses of water a day”.

He searched the electronic databases. He went through older literature that the databases do not cover. He consulted specialists in thirst and drinking. Then he wrote the sentence up front:

“No scientific studies were found in support of 8 × 8.” [1]

What he found instead is a better story than the absence.

The sentences that got dropped

Valtin reports two suggested origins — and we are going to be careful with the hedging here, because he was. His own section heading is “Possible origin of 8 × 8”, and this one reached him second-hand: according to one correspondent, another had suggested it. [1]

The suggestion is that it traces to this, from 1945:

Read the third sentence again, and then the two after it.

About 10½ cups (2.5 liters) a day — already more than eight 8-ounce glasses — and in the Board’s view most of it already in your food. Work or hot weather could push that total to somewhere between 21 and 55 cups (5 to 13 liters). And for the drinking itself, water was to be allowed ad libitum, Latin for as much as you want, because thirst is usually an adequate guide, except for infants and sick people [3]. (We convert at 1 US cup = 236.6 mL; an 8-ounce glass is one cup.)

That food line has not aged well. By its 1953 edition the Board itself had softened “most” to “much” [4]. US survey data put food at about a fifth of the water people take in [5], Europe’s food-safety agency assumes 20 to 30 percent [6], and the 2022 turnover study calls the share an unsolved question, with surveys ranging from 20 to 50 percent and probably undercounting food [2]. Nobody has shown that most of it comes from food. What survives is the point that matters here: the 1945 figure was for all the water you take in, food included, not a target for what you pour.

The suggestion, as Valtin reports it, is that the food sentence was not heeded, and a figure for total water became a target for drinking [1]. Even the arithmetic is loose: 2.5 liters is about 10½ eight-ounce glasses, not eight. And the retelling that reached Valtin had already dropped the last two sentences. It came from a 1994 newspaper feature, which quoted the first three and said “somehow, the final sentence was lost in the translation” [7]. It was not the final sentence. The final sentence is the one about thirst. Nobody has proved that is how it happened, and we are not going to tell you they have. But the 1945 text says what it says. It does not say “drink”. It says thirst.

An obituary said six. His 1974 book said 6 to 8.

This is our favourite detail in the paper.

After the comprehensive search, the nearest thing Valtin could find to an original recommendation was an obituary of a nutritionist, which described him as “an early champion of drinking at least six glasses of water a day.” [1]

Six. In an obituary. Then a former colleague of his went looking, and found a 1974 book on nutrition he had co-written. Valtin quotes the passage: water intake is “usually well regulated by various physiological mechanisms, but for the average adult, somewhere around 6 to 8 glasses per 24 hours and this can be in the form of coffee, tea, milk, soft drinks, beer, etc.” [1] Valtin calls it an “apparently offhand comment”, thinks it conceivable the rule began there, and notes that not one of the many sources he read ever cited it [1].

And to be clear, since the man is dead and did nothing wrong: what he wrote was somewhere around six to eight, with coffee, tea, milk, soft drinks and beer all counting. What went around the world was at least eight, of water, with the coffee struck off. Valtin calls the difference “huge” [1].

Coffee counts

The rule usually arrives with a rider — that tea, coffee and anything caffeinated do not count, and may even work against you.

Valtin found the published evidence does not support that either. Caffeinated drinks may be counted toward the daily total, and the list of things that count is long: “tap water and bottled water, coffee, tea, soft drinks, milk, juices, and possibly even beer in moderation.” [1]

Which makes the daily total a much easier target than it sounds, because you have been hitting most of it without counting.

One thing worth knowing about where that comes from. The trial Valtin leans on, 18 men drinking water, colas and coffee, was supported by a grant from The Coca-Cola Company [8]. The US Institute of Medicine, weighing it alongside older studies in 2004, reached the same place: caffeinated drinks “appear to contribute to the daily total water intake similar to that contributed by noncaffeinated beverages” [5].

Please do not take the wrong message from this

Valtin put this in his own paper twice, and we are putting it here in bold, because it is the part that gets lost when this research becomes a headline.

His conclusion applies to healthy adults, in a temperate climate, who are largely sedentary. [1] In the same breath he writes that intakes equal to or greater than eight glasses are advisable for treating or preventing certain diseases, and “certainly are called for” during vigorous work or exercise, especially in hot climates. [1]

So: the specific number has nothing behind it. Drinking water does. Those are not the same statement. Nothing here says to drink less than your thirst asks for. It says you do not need to force a number, and Valtin gives a whole section to the harm that forcing a lot of fluid can do, up to blood sodium diluted dangerously low, which doctors call hyponatremia [1].

What actually determines how much you need

In 2022, researchers measured it properly. They tracked water through the bodies of 5,604 people, from eight days old to 96, across 23 countries, using isotope methods rather than asking people what they drank. [2]

What predicted how much water a body used? Age. Body size. Body composition. Physical activity. Whether you are an athlete. Whether you are pregnant. And your environment — latitude, altitude, air temperature, humidity. Even how developed the country was showed up: people in countries low on the human development index turned over more water. [2]

Faced with that, the researchers did not publish a single target. They published equations, with worked examples of how far apart two people can be [2]. A non-athletic 20-year-old man of 154 lb (70 kg), at sea level in a highly developed country, at 50°F (10°C) and 50% humidity, is predicted to turn over about 13½ cups (3.2 L) of water a day; a non-athletic woman of 132 lb (60 kg) in the same place, about 11½ cups (2.7 L). Make them athletes at about 6,600 ft (2,000 m), in 86°F (30°C) air at 90% humidity, and the figures become about 31 and 29 cups (7.3 and 6.8 L) [2]. There is no single answer to give — the honest output of measuring 5,604 people is a formula with your circumstances as inputs. They said so plainly, too: the common suggestion to drink eight 8-ounce glasses a day “is not backed by objective evidence” [2].

One precision note, since it matters: what they measured was water turnover, the total moving through you each day. That includes water from food and water made by your own metabolism. It is not a drinking target either.

So does anyone publish a number?

Yes, and it pays to know what kind of number it is. In 2004 the Food and Nutrition Board — the same body, by then part of the US Institute of Medicine — set what it calls adequate intakes, meaning an amount that should meet the needs of almost everyone in a group: about 15½ cups (3.7 L) a day of total water for men and about 11½ cups (2.7 L) for women, counting food and every kind of drink, the same from age 19 into old age, and more in pregnancy and breastfeeding [5]. Drinks made up about 13 cups (3.0 L) and 9 cups (2.2 L) of that, around 81 percent [5]. In 2010 Europe’s food-safety agency set its own, lower figures for total water: about 10½ cups (2.5 L) a day for men and 8½ cups (2.0 L) for women, the same for older adults, plus about 1¼ cups (300 mL) in pregnancy and about 3 cups (700 mL) when breastfeeding, and only at moderate temperatures and moderate activity [6].

Now look at what those numbers are. The US figures are median intakes, the middle of the range, in a national survey — not a measured need — and the report says so in as many words: the adequate intake “does not represent a requirement”, and a wide range of intakes is compatible with normal hydration [5]. Day to day, it says, thirst and drinks with meals keep people hydrated [5]. That is not far from the last sentence of 1945.

One group should not lean on thirst alone. Thirst dulls with age: the US report calls impaired thirst in older people a long-held clinical observation that studies confirm, and sets their figure at the younger adults’ intake for that reason [5]; the European opinion and the 2022 turnover study both flag a reduced thirst response in older adults [6] [2]. The 1945 Board made its own exceptions, for infants and sick people [3].

A note on how the paper ends

Valtin finished by pointing out that you cannot prove a negative — he could not prove no supporting study exists anywhere — and so he invited readers who knew of one to write in. [1]

That is a correction policy, published inside the paper, in 2002. We extend the same offer on everything here.

The verdict

“Eight glasses of water a day” is Unsupported. A comprehensive search, in its author’s own word, found no study behind it. Its origin has two documented candidates: a 1945 figure for all the water you take in, which the Board thought came mostly from food and which ended by saying thirst is usually an adequate guide; and a line in a 1974 book on nutrition, “somewhere around 6 to 8 glasses”, coffee, tea and beer included. Neither says what the rule says, and nobody has shown which, if either, it came from.

What replaces it is not a target. The reference bodies publish typical intakes, not requirements, and your own need moves with your size, your activity, the weather and whether you are pregnant — while a healthy adult with a working sense of thirst and a normal diet is already most of the way there.

If you are older, ill, pregnant, working hard or somewhere hot, that is a conversation with a clinician, not a rule of thumb — thirst gets less reliable with age. We read the research; we are not your doctor.

Related: does drinking water help you lose weight · lemon water · all the habits we have checked · how many steps a day · sauna benefits