How much water you should drink each day and why needs vary
The research does not support a single daily figure, and the measurements describe a range instead. Isotope tracking in 5,604 people aged from eight days to 96 years across 26 countries put the low end of daily water turnover at around one to one and a half litres and the high end near six, with a handful of individuals above ten (Yamada et al., 2022). Nothing in that distribution looks like a single daily target.
This explainer follows the measurements rather than the recommendations: what the isotope method actually records, how the day's water arrives and leaves, how wide the spread is between people and within the same person, how quickly the body's own control loop reacts, and where the eight glasses rule came from in the first place. Related work is collected in the research digest.
What the isotope studies actually measure
The quantity researchers can measure is not how much a person drinks but how much water moves through them. Water turnover is the total traffic, intake and loss together, and it is measured by giving labelled water and watching the isotope disappear from the body over days (Yamada et al., 2022). The method answers a physiological question precisely, and it is the reason the research literature talks about turnover while the popular literature talks about glasses.
The purpose of that work is also not personal advice. The authors frame it as the basis for anticipating how much drinking water and water-rich food populations will need, in a world where one person in three lacks access to safe drinking water, and they state the distinction outright: turnover is not the same as daily water intake, and intake from liquids and food amounts to roughly 85% of it (Yamada et al., 2022).
The two are not interchangeable. Turnover includes water that arrives in food and water made by the body during metabolism, and it includes losses through skin and breath that nobody counts. Drinking is a fraction of it, and in American survey data plain water accounted for 30% of total water intake in men and 34% in women, with the rest arriving in other drinks and in food (Rosinger & Herrick, 2016). A target expressed in glasses is therefore measuring only one part of the system, and not the largest one.
Body composition sets the scale of the whole exchange. Water makes up about 60% of body weight in infants, around 50% in older adults and about 42% in women at 60, the difference reflecting how much of the body is fat rather than lean tissue. The daily traffic relative to that pool also changes with age: infants turn over roughly a quarter of their body water each day, adults around a tenth, and in adults turnover runs at about a third of a litre for every megajoule of energy expended (Yamada et al., 2022). A number that scales with lean mass and with energy use cannot be constant across people.
Where the day's water comes from and where it goes
The accounting is simple in structure. Water leaves as urine, as vapour through the skin and the lungs, and in faeces; it arrives in drinks, in the moisture of food, and from metabolism itself (Yamada et al., 2022). Commonly used balance figures put faecal loss near a tenth of a litre and insensible loss near nine tenths of a litre, against roughly a litre from food and three tenths of a litre made by metabolism; applying those to the measured 24-hour urine volume of 69 students left average drinking at about 1.2 litres (Valtin, 2002).
Water associated with glycogen storage is a separate issue from daily water turnover and should not be used to infer how much a person needs to drink. The figure usually quoted for water bound to stored carbohydrate traces back to a single old experiment, which is examined separately in does glycogen hold three grams of water.
Urine is the part the body adjusts. Measured directly in 280 adults, 24-hour urine production averaged 2.2 litres in both sexes, but ranged from 0.6 to 4.9 litres in men and from 0.9 to 6.0 in women (Raman et al., 2004). The kidney is the valve in this system, adjusting how much water leaves the body as intake rises and falls.
The same measurements rearrange the usual accounting. Urine covered about two thirds of total water output rather than the half often assumed, and the water arriving in drinks and food covered only 80% to 85% of turnover, the remainder being made by metabolism or absorbed from humid air (Raman et al., 2004).
Measuring intake at the level of a population is harder than it sounds, and that is part of why the question stays open. A review of the field noted major gaps in how total fluid intake and hydration status are measured across populations, few long-term systematic interventions and no published long-term randomised trials, alongside a shift in what people drink towards beverages that carry calories with the water (Popkin et al., 2010). Survey estimates and isotope studies are therefore measuring different things, and neither was designed to produce a personal target.
How wide the measured spread is, and what explains it
Two datasets answer this with numbers. In 458 adults aged 40 to 79 living in temperate climates, water from drinks and food averaged 3.0 litres in men and 2.5 in women, with individual values running from 1.4 to 7.7 litres and from 1.2 to 4.6 (Raman et al., 2004). The authors then asked how much of that variation body measurements could account for, and the answer was blunt: age, weight, height and body mass index together explained less than 12%.
The larger study adds the environment. Across 26 countries, turnover tracked age, body size and composition, physical activity, athletic status, pregnancy, socioeconomic status, latitude, altitude, air temperature and humidity, and people in countries with a low human development index had higher turnover than those in high-index countries (Yamada et al., 2022).
What was measured | Reported value | Source |
Water turnover, extra in the third trimester | About 0.67 litres a day | Yamada et al. (2022) |
Water turnover, extra during lactation | About 0.26 litres a day | Yamada et al. (2022) |
Men in the sample exceeding ten litres a day | 9 of 1,875 | Yamada et al. (2022) |
Variation between people explained by the full model | 47.1% | Yamada et al. (2022) |
24-hour urine production, average | 2.2 litres | Raman et al. (2004) |
24-hour urine production, range in men | 0.6 to 4.9 litres | Raman et al. (2004) |
24-hour urine production, range in women | 0.9 to 6.0 litres | Raman et al. (2004) |
Note: these are measured turnover and output figures from study populations, not intake targets for an individual. Of the nine men above ten litres a day, four were athletes and four were forager-horticulturalists in Amazonian Ecuador, which is the kind of detail an average conceals. The prediction model accounted for 47.1% of the variation between people, far better than body size alone and still leaving more than half unexplained (Yamada et al., 2022).
Measurement does not produce a daily number for water. It produces a distribution several litres wide, and body size explains barely a tenth of it.
The same person is also not a constant, though they vary far less than the population does. Repeat measurements put the within-person coefficient of variation at 8% for water intake and 9% for 24-hour urine production (Raman et al., 2004). Age shifts the picture too: in national survey data, men aged 60 and over took in 2.92 litres of total water against 3.63 litres in those aged 40 to 59, with a similar gap in women (Rosinger & Herrick, 2016).
Group differences sit on top of that individual spread. Intake rose with physical activity level in both sexes in the national survey, and non-Hispanic white adults recorded higher daily water intake than non-Hispanic black and Hispanic adults (Rosinger & Herrick, 2016). The isotope study of American adults found the same direction in the oldest age band it could analyse, with lower intake and lower turnover in black participants than in white ones (Raman et al., 2004). Neither study treats those gaps as biological requirements; they are descriptions of what different groups actually consumed.
How the body's own control loop responds
Water balance is closely regulated in healthy adults, although illness, heavy exertion and heat can increase fluid losses, and some groups are at greater risk of dehydration. For exercise-specific context, sport nutrition and hydration explains why sweat losses, exercise duration and environmental conditions can change fluid needs.
A loss of about 1% of body water is usually compensated within 24 hours, and the trigger is a minute change in the concentration of the blood rather than any sensation of effort (Jéquier & Constant, 2010). Two separate responses do the work: the kidney changes how concentrated the urine is, and thirst changes intake.
Their thresholds are not the same, and the order matters. In the example Valtin cites, the osmotic threshold for thirst sits at about 294 units on the scale used for blood concentration, and later reviews put the average lower and the spread between people wide, while the threshold for the hormone that concentrates urine is slightly lower, so the kidney starts adjusting before thirst is felt (Valtin, 2002). The normal range of blood concentration runs from 280 to 296 on that scale, which means thirst begins while the value is still inside it, and dehydration as usually defined begins near 302.
That ordering is what makes the familiar warning backwards. A rise in blood concentration of under 2% can produce thirst, whereas dehydration is conventionally defined from a loss of 3% or more of body weight, which corresponds to a rise of at least 5% (Valtin, 2002). Thirst is an early signal in this system, not a late one.
Where the eight glasses rule came from
The rule has a traceable origin and no supporting trial behind it. A 1945 recommendation from the Food and Nutrition Board of the National Research Council gave a suitable allowance of about two and a half litres of water a day for adults, and the sentence immediately after it noted that most of that amount is already contained in prepared foods. The review that traced the history reports a suggestion that the second sentence was dropped and the first was read as an instruction to drink (Valtin, 2002).
The same review searched for evidence supporting the rule and reported finding no scientific studies in its favour, while surveys of food and fluid intake in thousands of healthy adults suggested such large volumes were not needed. It also addressed the accompanying claim that caffeinated drinks do not count: in a controlled comparison of 18 healthy men given equal volumes of different beverages, there were no significant differences in the hydration measures used (Valtin, 2002). The stricter version of the rule is therefore an addition to an already unsupported one.
Where higher intake has been associated with outcomes
Claims about particular kinds of water sit in the same category and deserve the same treatment, which is why it pays to look at the evidence behind alkaline water and blood lipids rather than at the marketing around it. That separate research concerns blood lipid outcomes, not how much water a person needs to drink each day.
The case against a universal rule is not a case that volume never matters. The same review that found no support for the eight glasses rule set out the places where higher intake has been associated with measured outcomes (Valtin, 2002).
One observational association discussed by Valtin (2002) concerned bladder cancer. In a ten-year study of nearly 48,000 men, higher fluid intake was associated with lower incidence, with an estimated 7% reduction in risk per additional cup. However, this finding did not establish causation or a universal drinking target. A systematic review of prospective cohort studies found inconsistent evidence across four studies of fluid intake and did not support a strong or consistent association with bladder cancer incidence (Miyajima et al., 2026).
The reviewer's own conclusion was that the evidence for forcing high intake on healthy adults in a temperate climate is weak at best, and that further data would be needed before recommending the rule universally even for preventing those conditions. He also made the opposite point concretely: modest increases in intake can produce severe water intoxication when the kidney's ability to excrete water is limited by sustained hormonal action, a situation he illustrated with a patient whose fluid intake had previously been governed safely by thirst alone (Valtin, 2002). The distribution has two edges, and both of them are measured.
What the measurements support and what they do not
Supported: water turnover varies several-fold between people, and anthropometric variables explain little of the variation (Raman et al., 2004). A substantial share of total water intake comes from food and beverages other than plain water (Rosinger & Herrick, 2016). Water-balance regulation can correct small deficits within a day, and the eight glasses rule has a documentary rather than experimental origin (Jéquier & Constant, 2010; Valtin, 2002).
Not supported in the way it is usually presented: urine colour as a readout of adequacy. A pooled analysis of 238 adults found that shifting 24-hour urine colour by two shades on an eight-colour scale corresponded to a change in total fluid intake of roughly 1.1 litres in either direction (Perrier et al., 2016). Colour tracks a change in intake, and tracks it coarsely; it is not a measurement of whether a given intake was enough.
Unsettled: the requirement itself. A recent review states plainly that human water requirements remain undefined, because of the complexity of the regulatory network and the size of the differences between individuals, and proposes assessing them from the intensity of the body's own hormonal response rather than from survey averages (Armstrong & Johnson, 2018).
Water intake guidelines nevertheless provide population reference values, not exact prescriptions for every person. For example, the European Food Safety Authority sets adequate total-water intakes at 2.0 litres per day for adult women and 2.5 litres for adult men under moderate temperature and activity conditions, including water from food and all beverages (EFSA NDA Panel, 2010).
An earlier review of the field noted major gaps in measuring intake and hydration at population level and the absence of long-term randomised trials (Popkin et al., 2010). The honest answer to the question in the title is that the measurements describe a wide distribution and a fast correction loop, and that no study has produced the single number the question expects.
Frequently asked questions
Do tea and coffee count towards the total?
The controlled comparison cited most often found no significant differences in hydration measures between equal volumes of water and of caffeinated and non-caffeinated drinks in 18 healthy men. The claim that caffeinated drinks must be excluded was attached to the eight glasses rule rather than derived from a trial. (Valtin, 2002).
How much of the day's water arrives in food?
Enough to change the arithmetic. Balance figures put the moisture in food at roughly a litre a day with a further small amount produced by metabolism, and in American survey data plain water accounted for only about a third of total water intake, the rest coming from other drinks and from food. (Valtin, 2002; Rosinger and Herrick, 2016).
Is it true that thirst arrives too late?
The measurements say the opposite. Thirst begins after a rise in blood concentration of less than 2%, while dehydration is conventionally defined from a loss of 3% or more of body weight, which corresponds to a much larger rise. The kidney starts adjusting urine before thirst is even felt. (Valtin, 2002).
Can urine colour show whether intake is adequate?
It shows change rather than adequacy. In a pooled analysis of 238 adults, moving urine colour by two shades on an eight-colour scale corresponded to a difference of about 1.1 litres in daily fluid intake, so the scale is coarse and says nothing about whether the starting point was sufficient. (Perrier et al., 2016).
Does the amount change with age?
Measured intake falls in later life. In survey data men aged 60 and over took in about 2.9 litres of total water against about 3.6 litres among those aged 40 to 59, with a similar difference in women. Isotope studies record lower turnover in older adults once body composition is accounted for (Rosinger and Herrick, 2016; Yamada et al., 2022).
Sources
- Valtin H. Drink at least eight glasses of water a day. Really? Is there scientific evidence for 8 x 8? American Journal of Physiology: Regulatory, Integrative and Comparative Physiology. 2002;283(5):R993–R1004. DOI 10.1152/ajpregu.00365.2002.
- Raman A, Schoeller DA, Subar AF, et al. Water turnover in 458 American adults 40–79 yr of age. American Journal of Physiology: Renal Physiology. 2004;286(2):F394–F401. DOI 10.1152/ajprenal.00295.2003.
- Yamada Y, Zhang X, Henderson MET, et al. Variation in human water turnover associated with environmental and lifestyle factors. Science. 2022;378(6622):909–915. DOI 10.1126/science.abm8668.
- Jéquier E, Constant F. Water as an essential nutrient: the physiological basis of hydration. European Journal of Clinical Nutrition. 2010;64(2):115–123. DOI 10.1038/ejcn.2009.111.
- Popkin BM, D'Anci KE, Rosenberg IH. Water, hydration, and health. Nutrition Reviews. 2010;68(8):439–458. DOI 10.1111/j.1753-4887.2010.00304.x.
- Rosinger A, Herrick K. Daily water intake among U.S. men and women, 2009–2012. NCHS Data Brief No. 242. Hyattsville, MD: National Center for Health Statistics; 2016.
- Perrier ET, Johnson EC, McKenzie AL, Ellis LA, Armstrong LE. Urine colour change as an indicator of change in daily water intake: a quantitative analysis. European Journal of Nutrition. 2016;55(5):1943–1949. DOI 10.1007/s00394-015-1010-2.
- Armstrong LE, Johnson EC. Water intake, water balance, and the elusive daily water requirement. Nutrients. 2018;10(12):1928. DOI 10.3390/nu10121928.
- EFSA Panel on Dietetic Products, Nutrition and Allergies (NDA). Scientific Opinion on Dietary Reference Values for water. EFSA Journal. 2010;8(3):1459. DOI: 10.2903/j.efsa.2010.1459.
- Miyajima K, Miszczyk M, Roessler N, et al. Diet and risk of bladder cancer: a systematic review of epidemiological evidence. BJU International. 23 August 2026. DOI: 10.1111/bju.70435.
Related
Does Glycogen Hold 3 g of Water? What the 1941 Rat Study Found
Alkaline Water and Blood Lipids: What a 2009 Rat Study Tested
Sport Nutrition: Energy, Fuel and Athletic Performance
Impostor syndrome and why some people doubt their success
Is coffee good for you and what research says about its health effects
Travel in Later Life: What an Eight-Person Study Found
Adult Literacy Classes in Kenya: What a 1980s Study Found