Recent studies have shown that potentially fatal hyponatremia can develop during prolonged exercise. To determine the incidence of hyponatremia in athletes competing in ultradistance events, we measured serum sodium levels in 315 of 626 (50%) runners who were treated for collapse after two 90 km ultramarathon footraces (total starters 20,335; total finishers 18,031) and in 101 of 147 (69%) finishers in a 186 km ultratriathlon. In both races the athletes drank fluids with low sodium chloride content (less than 6.8 mmol.l-1). Hyponatremia (serum sodium level less than 130 mmol.l-1) was identified in 27 of 315 (9%) collapsed runners in the 90 km races and in none of the triathletes. In response to diuretic therapy, the runner with the most severe hyponatremia (serum sodium level = 112 mmol.l-1) excreted in excess of 7.5 l dilute urine during the first 17 h of hospitalization. These data suggest that, although symptomatic hyponatremia occurs in less than 0.3% of competitors during prolonged exercise even when they ingest little sodium chloride, it is found in a significant proportion (9%) of collapsed runners. A regulated contraction of the extracellular fluid volume would explain why the majority of athletes maintain normal serum sodium levels even though they develop a significant sodium chloride deficit during prolonged exercise. Alternatively, sodium chloride losses during prolonged exercise may be substantially less than are currently believed. Physicians treating collapsed ultradistance athletes need to be aware that as many as 10% or more of such patients may be hyponatremic.