Drinking and Training: What Water, Salt and Performance Really Connect

Direct answer: water is an ergogenic substance, not background noise

Losing more than 2 percent of bodyweight to sweat cuts endurance by roughly 15 to 20 percent, strength by 5 to 10 percent, and concentration measurably (Goulet 2017, meta-analysis of 64 trials). At 80 kg that's 1.6 litres of sweat, easy to hit in a 90-minute indoor session. On the other side, over-drinking is real and dangerous: exercise-associated hyponatraemia has killed athletes (Brusati & Cintoni 2025, J Endocrinol Invest).

The right answer sits in a narrow band: drink what you lose, plus sodium when the loss is big. Not more.

What dehydration actually does

Just 1 percent body-water loss raises heart rate at a given workload by about 5 beats per minute. At 2 percent, stroke volume reserve drops, sweating gets less efficient, and core temperature climbs faster. Past 3 percent, reaction time and decision tasks degrade as much as after 24 hours without sleep (Wittbrodt & Millard-Stafford 2018).

In strength training the effect is smaller but real: bench press and squat 1RM drop by 5 to 8 percent at 2 to 3 percent dehydration. That's the gap between a good session and a mediocre one. In endurance sport the hit is sharper: 5K run time slows by roughly 25 seconds at 2 percent dehydration.

How much you actually need per day

Standard guidance: about 35 ml per kg of bodyweight per day baseline. At 75 kg that's 2.6 litres total, of which roughly 0.7 litres come from food. That leaves 1.9 litres to drink, plus the training-dependent top-up.

Four espressos a day don't add to your need. Caffeine at moderate doses isn't diuretic, that myth keeps resurfacing. Alcohol is, costing you roughly 10 ml extra water per gram of alcohol.

How to measure your sweat rate

Weigh yourself nude before and after a session on the same scale. Difference in grams = sweat loss in millilitres (1 g ≈ 1 ml). Add the fluid you drank during. Example: 80.5 kg before, 79.2 kg after, 500 ml drunk in 60 min. Sweat rate: 1.3 kg lost + 500 ml drunk = 1800 ml/h. In a 2-hour session you'd need to replace 3600 ml, not 1800.

Electrolytes: sodium is the only one that matters

Sweat carries mainly sodium and chloride, very little potassium, almost no magnesium or calcium. Concentrations vary wildly between people, from 200 to 1800 mg sodium per litre of sweat. If you finish training with white salt streaks on your shirt or cap, you're a salty sweater and need more.

For sessions under 60 minutes, you don't need electrolytes. Plain water is fine. For longer sessions, heat, or multiple sweat sessions in one day, 300 to 800 mg sodium per litre helps. Sports drinks supply that (around 400 to 600 mg/L), but a homemade mix of water plus a small pinch of salt plus some sugar does the same job (Maughan 2012, ACSM).

Magnesium tablets against cramps are popular, but the evidence is thin. Cramps tend to track local muscle fatigue and neural excitability more than magnesium status, which is rarely clinically low on a normal diet.

Hyponatraemia: the under-discussed risk

Drink too much pure water without sodium and your serum sodium drops below 135 mmol/L. Symptoms: headache, nausea, later confusion and seizures. The US military logged 134 cases in 2024 alone, a clear jump from previous years (Bates 2025). Marathon data shows 8 to 13 percent of finishers run hyponatraemic, almost always from over-drinking.

Rule of thumb: never drink more than you sweat. If you're heavier after a session than before, you drank too much.

What's still open

Pee-colour charts (8-step scale from clear to dark yellow) are useful as a hint but imprecise. B-vitamin supplements turn urine neon, some foods do too. Specific gravity (USG) on morning urine is more accurate but impractical day-to-day.

How much sodium an individual ideally needs in a sports drink depends more on genetics than on sport or climate. Wearable sweat sensors promise personalisation but aren't reliable yet (Frontiers 2025).

Short version: drink when you're thirsty, salt when you sweat long. That covers 95 percent of cases better than any complicated plan.