Almost everything you have been told about drinking on a run traces back to a sentence written in 1996 that the people who wrote it have since walked away from. Here is what the evidence actually supports — and the one mistake that has killed runners.
Start with the thing that can kill you
Drinking too much water during a long run can be fatal. This is not a theoretical risk. The international consensus panel on the condition counts at least fourteen deaths in athletes since 1981, across marathons, a canoe race, hiking, military and police training, and school sport.
The condition is exercise-associated hyponatraemia — blood sodium falling below 135 mmol/L. At the 2005 Boston Marathon, researchers took blood from 488 finishers. Thirteen per cent were hyponatraemic. Three runners in five hundred were below 120 mmol/L, which is the range where people have seizures and stop breathing.
What predicted it was not the weather, and not what people drank. It was gaining weight during the race — odds ratio 4.2 (95% CI 2.2 to 8.2) — and finishing in over four hours — odds ratio 7.4 (95% CI 2.9 to 23.1). Slower runners have more time to drink and more aid stations to drink at.
Read that first finding again, because it is the most useful rule on this page. If you weigh more after a run than before it, you drank too much. There is no situation in which that is correct.
One more finding from Boston, because it kills the most common workaround: the composition of what people drank made no difference. Sports drink, water, salt tablets — none of it separated the hyponatraemic runners from the rest.
Salt tablets do not protect you
This deserves its own section because the belief is so widespread and the evidence against it is so direct.
At the Western States 161 km ultramarathon, researchers surveyed runners and took post-race bloods. Ninety-four per cent were taking sodium supplements. Hyponatraemia still occurred in 6.6% of finishers — and, the key sentence, there was no difference in sodium intake between the runners who became hyponatraemic and those who did not.
There is a published case of a runner who drank 9 to 10 litres of water and took over 6,500 mg of sodium during a 17-hour run. His blood sodium finished at 122 mmol/L. The salt did not save him, because the problem was never a shortage of salt. It was an excess of water.
The consensus statement puts it precisely: sodium supplementation can slow the fall in blood sodium when you are drinking roughly what you are sweating, but it “cannot prevent EAH in the setting of a persistent excessive fluid intake.”
Where “drink before you are thirsty” came from
The 1996 American College of Sports Medicine position stand told athletes to “start drinking early and at regular intervals in an attempt to consume fluids at a rate sufficient to replace all the water lost through sweating… or consume the maximal amount that can be tolerated.”
That is the origin of the advice, and it is the advice that produced the deaths. The 2007 revision quietly replaced it with something much weaker — prevent losses greater than 2% of body weight — without much fanfare about the change.
Commercial money sits on both sides of this argument, and you should know it. A 2012 systematic assessment of sports-performance products found 52.8% of websites making performance claims cited no evidence at all, and of the studies that were cited, 84% were at high risk of bias and only 3 of 74 were high quality. It is equally true that the 2015 consensus conference which produced the “drink to thirst” guidance was itself sponsored by a commercial fitness company, and that several sweat-research authors are employed by a sports-drink manufacturer. Be sceptical in both directions.
Does being a bit dehydrated actually slow you down?
Less than you would think, and the famous 2% threshold rests on less than its ubiquity suggests.
That number traces to a 1992 study of eight cyclists pedalling at a fixed power output. It measured how hot they got and how hard their hearts worked. It did not measure performance at all — at a fixed workload there is no performance to measure.
When people tested it properly, the effect largely disappeared:
- The only running study. Ten trained runners ran a half marathon at 30°C twice — once drinking to thirst (384 mL/h, finishing 3.1% down on body weight), once on a programmed schedule (1,380 mL/h, 1.3% down). Finish times: 89.8 minutes versus 89.6 minutes. No difference. The programmed drinking did keep them slightly cooler with a lower heart rate — real effects, just not ones that showed on the clock.
- Blinded, with a drip. Ten cyclists were dehydrated by 3%, then given intravenous saline to restore them to 0%, 2% or 3% down without knowing which. Over 25 km in 33°C, core temperature was higher in the 3% condition after 17 km, and “no other differences between trials were shown.”
Two honest caveats. First, this literature is thin: the case against the 2% rule rests on roughly eighty subjects, almost all cyclists. Second, and more relevant to us, all of it was done in dry heat — 10 to 40% humidity. Gurugram in August is not that, and no blinded trial has been run in humid heat. Do not let anyone, including us, tell you the question is settled for our conditions.
What actually happens to a runner in Indian heat
There is no published research on hydration in Indian runners. None — the search returns nothing. The closest usable evidence is a study of 166 recreational runners in Thailand, running in 29.6°C and 70% humidity, a fair stand-in for a Gurugram morning outside winter.
- Sweat rate: 1.3 L/h in men, 0.9 L/h in women
- Sweat sodium concentration ranged from 11 to 80 mmol/L between individuals — a sevenfold spread
- Their conclusion: the standard fluid guidelines “are applicable to tropical native runners”
Now set that against a second number. The stomach can only empty so fast. In cyclists drinking as much as they could tolerate, the ceiling was about 1.1 litres per hour — and even then they reported feeling unpleasantly full. Running empties the stomach more slowly than cycling and causes more gut trouble.
So: sweating around 1.3 L/h, able to absorb perhaps 1.0 L/h at best and realistically much less while actually moving. In our climate you often cannot replace what you lose, however disciplined you are. Finishing a long summer run down a kilogram or two is arithmetic, not a failure of willpower. Plan around it instead of fighting it.
Cramp is not a hydration problem
The cleanest test took ten men and dehydrated them by 4.7% of body mass — nearly four litres of fluid and four grams of sodium. Then they measured how easily a muscle could be made to cramp.
Cramp threshold frequency: 15 ± 5 Hz when hydrated, 13 ± 6 Hz when dehydrated (p = 0.12). No difference. Serious dehydration with real electrolyte loss did not make muscles more likely to cramp.
The leading review notes that evidence for the dehydration and electrolyte theories amounts to “anecdotal clinical observations, case series totalling 18 cases, and one small (n = 10) case-control study,” while “four prospective cohort studies do not support these hypotheses.” The better-supported explanation is altered neuromuscular control — a fatigued muscle misfiring — though the honest verdict is that “strong experimental evidence for either theory is lacking.”
This matters more than it sounds. Three American high-school footballers died of hyponatraemia between 2008 and 2014 after being encouraged to drink heavily to prevent cramp. Bad cramp advice has killed people.
What we suggest
Drink to thirst. This is the consensus recommendation and we follow it. Know what it rests on: the panel graded it 1C — a strong recommendation on low-quality evidence — and their own document lists cases of serious hyponatraemia in people who were drinking to thirst. It is the best default available. It is not a guarantee.
Carry something on anything over an hour in the heat. Not to hit a number, but because thirst can only guide you if there is something to drink.
Never drink to a schedule that ignores how you feel, and never drink so much that you finish heavier than you started.
Learn your own sweat rate once. Weigh yourself before a run and again after, towelled dry. Add back whatever you drank. The difference over the hours you ran is roughly your sweat rate. Two things to know: it is accurate to about ±10%, not better, and some of that loss is fuel you burned rather than water — roughly 0.23 g per kilocalorie — so a long run should show some weight loss that is not dehydration. Repeat it across a few different days; the number moves.
Ignore urine colour charts as a spot check. The research that validated urine colour used controlled first-morning samples across a whole protocol. Glancing at one sample at a random time of day is a different thing, and the methodological reviews say the practice “can be confounded by fluid intake, diet, and exercise… leading to false positive/negative assessments.” When researchers built a model combining body weight, urine and thirst, body weight carried 58% of the predictive weight and thirst 26%. Urine colour was the weakest of the three.
Do not bother pre-loading. A 2025 trial had thirteen trained runners hyperhydrate before a field half marathon in warm conditions. It worked physiologically — an extra 1,189 mL retained (95% CI 987 to 1,391), plasma volume up 9.0%. There was no difference in performance, gut symptoms, core temperature or heart rate.
What about carbohydrate?
A separate question from fluid, and the answer depends almost entirely on how long you are out.
Under an hour: nothing needed. A carbohydrate mouth rinse — swill and spit — improved one-hour time-trial performance with no change in heart rate or perceived effort, which tells you the effect is in the brain rather than the muscle. For a 10K, fuel is essentially irrelevant.
Two to three hours: a single carbohydrate source can be absorbed at up to about 60 g/h. Beyond that the gut, not the muscle, is the limit.
Longer: around 90 g/h becomes possible, but only with a glucose-and-fructose mix, which uses two different transporters. Studies consistently show 40–55% higher oxidation rates for mixed sources over glucose alone.
One caveat for us: the same researchers measured this in the heat and found absorption was lower — 1.14 g/min at 32°C against 1.26–1.70 g/min in cool conditions. Assume you will handle less here than European lab numbers suggest. And every one of these studies measured how fast fuel was oxidised, not how fast anyone ran. That step is an inference.
You may see 120 g/h quoted from professional cycling. That is reported practice, not trial evidence, and it is not a target for amateur runners.
The short version
- Drink when you are thirsty. Carry fluid on long hot runs so that you can.
- Never finish heavier than you started. That is the one hard rule.
- Salt tablets do not prevent hyponatraemia and do not prevent cramp.
- In our heat you cannot fully replace what you sweat. Accept the deficit rather than chasing it.
- Finishing a long run a kilogram or two down is normal and will not have slowed you measurably.
- Under an hour you need nothing. Over two, think about carbohydrate before you think about salt.
If you feel confused, unusually nauseated, get a headache, or your hands and face look puffy during or after a long run — stop drinking and get medical help. Those are the warning signs of hyponatraemia, and drinking more will make it worse.
References
- Hew-Butler T, et al. Statement of the Third International Exercise-Associated Hyponatremia Consensus Development Conference. Clin J Sport Med, 2015;25(4):303–20. doi:10.1097/JSM.0000000000000221
- Almond CSD, et al. Hyponatremia among runners in the Boston Marathon. N Engl J Med, 2005;352(15):1550–6. doi:10.1056/NEJMoa043901
- Hoffman MD, Stuempfle KJ. Sodium supplementation and exercise-associated hyponatremia during prolonged exercise. Med Sci Sports Exerc, 2015;47(9):1781–7. doi:10.1249/MSS.0000000000000599
- Dion T, et al. Half-marathon running performance is not improved by a rate of fluid intake above that dictated by thirst sensation. Eur J Appl Physiol, 2013;113(12):3011–20. doi:10.1007/s00421-013-2730-8
- Wall BA, et al. Current hydration guidelines are erroneous: dehydration does not impair exercise performance in the heat. Br J Sports Med, 2015;49(16):1077–83. doi:10.1136/bjsports-2013-092417
- Montain SJ, Coyle EF. Influence of graded dehydration on hyperthermia and cardiovascular drift during exercise. J Appl Physiol, 1992;73(4):1340–50.
- Surapongchai J, et al. Hydration status, fluid intake, sweat rate, and sweat sodium concentration in recreational tropical native runners. Nutrients, 2021;13(4):1374. doi:10.3390/nu13041374
- Braulick KW, et al. Significant and serious dehydration does not affect skeletal muscle cramp threshold frequency. Br J Sports Med, 2013;47(11):710–4. doi:10.1136/bjsports-2012-091501
- Schwellnus MP. Cause of exercise associated muscle cramps (EAMC). Br J Sports Med, 2009;43(6):401–8. doi:10.1136/bjsm.2008.050401
- Baker LB. Sweating rate and sweat sodium concentration in athletes. Sports Med, 2017;47(Suppl 1):111–28. doi:10.1007/s40279-017-0691-5
- Jeukendrup A. A step towards personalized sports nutrition: carbohydrate intake during exercise. Sports Med, 2014;44(Suppl 1):S25–33. doi:10.1007/s40279-014-0148-z
- Jardine WT, et al. The addition of glycerol and sodium chloride to a hyperhydration protocol does not improve half-marathon time-trial performance. Int J Sports Physiol Perform, 2025;20(6):770–8. doi:10.1123/ijspp.2024-0268
- Cheuvront SN, Kenefick RW, Zambraski EJ. Spot urine concentrations should not be used for hydration assessment. Int J Sport Nutr Exerc Metab, 2015;25(3):293–7. doi:10.1123/ijsnem.2014-0138
- Heneghan C, et al. The evidence underpinning sports performance products: a systematic assessment. BMJ Open, 2012;2(4):e001702. doi:10.1136/bmjopen-2012-001702