Drinking to thirst or to a plan: what ultra data show

In a 161 km ultramarathon in temperatures up to 39.0°C, 67.0% of 383 starters drank to thirst for at least one race segment, and weight loss did not differ between drinking strategies. A 2025 narrative review of six studies concluded that neither thirst-driven nor programmed drinking seems universally superior.

What did the 161 km ultramarathon study find?

Hoffman and Stuempfle, Research in Sports Medicine 2014 weighed 383 starters before, during and after the race and asked them afterwards about drinking and sodium use in four race segments. Drinking to thirst was the most common strategy. Mean weight loss ranged from 2.0 to 3.1% and neither its extent nor its pattern differed between strategy groups.

Among the top 10 finishers, half had lost more than 2% of starting body weight by 90 km. The authors concluded that weight loss above 2% does not necessarily harm performance in ultra-endurance events with high thermal stress.

This was observational. Runners chose their own strategy, so the study cannot show that drinking to thirst caused any outcome, and it covers one hot race.

What does the wider review say about programmed drinking?

Wierick et al., Nutrients 2025 synthesised six studies, five from the field and one from the laboratory. Programmed intake means fluid volumes set from an estimated sweat rate. In controlled laboratory trials, it supported plasma volume, reduced cardiovascular strain and improved performance during prolonged exercise.

In field studies of experienced ultra-runners, thirst-driven drinking showed no increased incidence of exercise-associated hyponatremia (low blood sodium) and no measurable performance impairment, even with moderate body mass loss. The review notes that hyponatremia risk is well documented, especially when fluid intake exceeds sweat loss. It also says thirst-driven intake may underperform in people with high sweat rates or impaired thirst perception.

The authors suggest a hybrid model that combines individual sweat testing, environmental context and internal cues. Individual testing is the same logic as comparing a result with a personal baseline rather than a population range.

The bottom line

• In a 161 km race, 67.0% of 383 starters drank to thirst, and weight loss (mean 2.0 to 3.1%) did not differ between drinking strategies.
• A 2025 review of six studies found no increased hyponatremia with thirst-driven drinking in field studies of experienced runners, and better-controlled laboratory results favouring programmed intake.
• The evidence is small and mostly observational, and it concerns experienced ultra-runners, not all athletes.