How Much Water Should You Drink? Past the Eight-Glasses Myth

Most healthy adults need somewhere in the region of two to three litres of total water a day, and the large majority reach it without counting anything. That is the honest version of the answer. The figure moves with body size, activity, air temperature and what you eat, and the mechanism that tracks all four of those variables simultaneously is thirst, not arithmetic.

None of which makes a fluid target worthless. It makes it a planning figure, useful mainly for people whose thirst signal is blunted by age or medication, whose training load is heavy enough to outrun it, or who have simply never noticed how little they drink. What follows is where the eight-glasses rule came from, what the reference bodies publish, what a body-weight formula adds, and which signals from your own body outrank all of it.

The eight glasses have no source

Heinz Valtin, a kidney physiologist at Dartmouth, went looking for the origin of the advice and published what he found in the American Journal of Physiology (2002). He could not locate it. No trial, no consensus statement, no cohort study underpins the eight-by-eight rule. The closest candidate he identified was a 1945 Food and Nutrition Board recommendation of roughly one millilitre of water per kilocalorie consumed, which for a 2,000 kcal diet works out at about two litres. The same document noted that much of this quantity arrives in food. That qualifying sentence dropped out of the retelling, and a figure describing total water intake became a figure describing glasses of water to be drunk on top of everything else.

Valtin's review also examined the corollary claim that thirst arrives too late to be useful, and found it unsupported in healthy people. Osmoreceptors in the hypothalamus respond to changes in plasma osmolality of around one to two percent, a smaller deviation than the level at which performance or cognition measurably suffers. The alarm sounds before the damage, which is what an alarm is for.

What the reference bodies actually publish

Two bodies have set formal reference values, and both frame them more cautiously than the internet does.

The EFSA Panel on Dietetic Products, Nutrition and Allergies, in EFSA Journal 8:1459 (2010), set adequate intakes for total water at 2.0 litres a day for adult women and 2.5 litres for adult men, under temperate conditions and moderate activity. Total water means everything: drinks, the water in food, and the small amount produced by metabolism itself.

The Institute of Medicine's Dietary Reference Intakes for Water, Potassium, Sodium, Chloride and Sulfate (2005) arrived at higher figures, 2.7 litres for women and 3.7 litres for men, of which roughly 80% came from beverages in the populations surveyed. The gap between the two sets of numbers is not a scientific dispute so much as a methodological one. The IOM panel could not identify an intake at which health outcomes improved, so instead of deriving a requirement it reported the median total intake of apparently healthy people and called that adequate. The report states plainly that the vast majority of healthy people meet their needs by letting thirst guide them.

Read those documents together and the practical message is narrower than either headline number suggests. There is a broad band of intakes compatible with good health, the band shifts with circumstance, and no authority has demonstrated a benefit to sitting at the top of it.

What a body-weight formula adds

A per-kilogram figure is more useful than a fixed one for the obvious reason: a 55 kg adult and a 95 kg adult do not have the same requirement. Our water intake calculator starts from roughly 35 ml per kilogram of body weight, a convention drawn from clinical maintenance-fluid practice, then adds for exercise and climate.

For a 70 kg adult with no training and a temperate climate, that baseline comes to 2,450 ml of total water a day. Subtract the share that typically arrives in food, around 20% in a normal mixed diet, and the drinking target is 1,960 ml, or about eight 250 ml glasses.

The coincidence is worth sitting with. A formula built from body weight and a food-water correction lands almost exactly on the folk rule, for one particular body weight, in one particular climate, at rest. Move any of those inputs and the agreement breaks. The same calculation for a 60 kg adult gives 2,100 ml total and roughly seven glasses; for a 90 kg adult it gives 3,150 ml and about ten. The eight-glasses rule is not wrong so much as it is a single point on a line, presented as the whole line.

Figures are for a temperate climate with no deliberate exercise. They are estimates of a comfortable intake, not thresholds, and the error band around them is wide.

Food is a fluid source, and so is coffee

The assumption that only plain water counts is the single largest reason people believe they are behind. Fruit and vegetables run from 80% to over 95% water by weight. Soup, yoghurt, milk, porridge, rice and pasta all carry substantial water. The IOM's own survey data attributed about a fifth of total intake to food, and for someone eating a diet heavy in produce the share is higher still.

Caffeinated drinks count too, which contradicts a durable piece of folk wisdom. Caffeine has a mild diuretic action at high acute doses in people unaccustomed to it, but the effect does not produce net fluid loss at habitual intakes. Killer, Blannin and Jeukendrup tested this directly in PLOS ONE (2014), comparing four mugs of coffee a day against an equal volume of water in habitual coffee drinkers, and found no difference in any measure of hydration status across the two conditions. The fluid in the cup more than covers the small increase in urine output. Alcohol is the genuine exception: it suppresses vasopressin and does produce a net negative fluid balance, which is a large part of what a hangover consists of.

The signals that beat any formula

Two observations, both free, carry more information about your fluid status than any calculation.

The first is thirst. In a healthy adult with intact kidneys and normal access to fluid, drinking to thirst maintains plasma osmolality within a remarkably tight range. The system has redundancy and it is fast. The exceptions are specific and worth knowing. Thirst sensitivity declines with age, so older adults can be genuinely dehydrated without feeling it. Some medications blunt the signal or alter fluid handling. And during hard exercise, particularly in heat, voluntary drinking commonly lags behind sweat loss, which is the situation formal fluid guidance exists to address.

The second is urine colour. A pale straw colour suggests adequate hydration; consistently dark, strongly smelling urine in small volumes suggests the opposite. Armstrong and colleagues validated colour charts against measured urine specific gravity and osmolality in the International Journal of Sport Nutrition (1998) and found the correlation good enough for field use. Two caveats: B vitamins from supplements turn urine bright yellow regardless of hydration, and the first void of the morning is normally concentrated in anyone. Judge across the day, not on one reading.

What neither signal supports is the pursuit of colourless urine. Clear urine all day means you are drinking beyond what your kidneys need to excrete, and the surplus is simply passed through.

When training changes the arithmetic

Exercise is where a target starts to earn its keep, because sweat rates vary by an order of magnitude between individuals and conditions, from under half a litre an hour to well over two.

The ACSM Position Stand on exercise and fluid replacement, published in Medicine and Science in Sports and Exercise (2007), recommends drinking during exercise to limit body-mass loss to less than about 2%, and gives roughly 400 to 800 ml per hour as a starting range for most athletes, to be individualised by measuring sweat rate. The calculator here adds 600 ml per hour of exercise, the midpoint of a comparable band, then adds a further 15% to 25% of baseline for hot conditions. For our 70 kg adult, an hour of training raises total water from 2,450 ml to 3,050 ml; ninety minutes of training in hot conditions takes it to about 3,720 ml.

Measuring your own sweat rate is straightforward and replaces all of this guesswork. Weigh yourself unclothed before and after an hour of training, note anything you drank, and each kilogram of body mass lost corresponds to roughly a litre of sweat. Do it in two or three different conditions and you will know your own range better than any equation can tell you. The calories burned calculator will give you a sense of session intensity from published MET values, which correlates loosely with sweat production, though heat and humidity matter more than intensity alone.

Sweat is not only water. Once sessions run long or conditions are hot, sodium losses start to matter and plain water stops being a complete replacement, which is the subject of our guide to electrolytes and hydration.

Where more helps, and where more harms

For most people, extra fluid beyond comfortable satiety does nothing measurable. Two exceptions have reasonable evidence behind them. Kidney stones are the clearest: higher urine volume reduces the concentration of stone-forming solutes, and increased fluid intake is a standard part of recurrence prevention, though anyone who has passed a stone should follow a clinician's specific target rather than a general one. Constipation is the second, with a narrower claim than usually made. In people who are genuinely under-hydrated, more fluid alongside adequate fibre improves stool consistency; in people already drinking adequately, adding more does little, and the fibre matters more than the volume, as covered in our guide to carbohydrates and fibre.

Overhydration receives far less attention than dehydration and has killed athletes. Exercise-associated hyponatraemia occurs when fluid intake exceeds losses to the point that blood sodium becomes diluted, and it has been documented most often in endurance events where participants drink heavily at every aid station over many hours. Almond and colleagues reported in the New England Journal of Medicine (2005) that a meaningful minority of Boston Marathon finishers crossed the line with low blood sodium, with the strongest predictors being substantial weight gain during the race and a longer finishing time.

Early symptoms, headache, nausea, confusion and bloating, overlap almost entirely with dehydration, which makes self-diagnosis during an event unreliable and dangerous. Gaining weight during prolonged exercise is the signal that intake has exceeded losses. Hyponatraemia and its management are clinical matters and nothing in this article should be used to self-treat suspected cases; the same applies to the fluid restrictions prescribed in heart failure, advanced kidney disease, cirrhosis and certain hormonal conditions, where a clinician's target overrides any calculator output without exception.

A workable approach

Start from your body weight rather than a fixed number of glasses. Add roughly half a litre to three-quarters of a litre for each hour of training, and more in heat. Count everything you drink, including coffee and tea, and remember that a fifth or so of your total arrives in food.

Then stop calculating and start observing. Drink to thirst, check that your urine is pale rather than dark across the day, and treat persistent headaches, dark urine or poor tolerance of heat as a prompt to drink more. If you train hard, weigh yourself before and after a session once or twice to learn your actual sweat rate. Those observations will serve you better across a year than any formula, including ours, which exists to give you a starting figure rather than a rule to obey.