This study examined the effect of the sodium content of drinks on the rehydration process after exercise. Six healthy male volunteers were dehydrated by a mean (SEM) of 1.9(0.0) % of body mass by intermittent cycle exercise in a warm (32 degrees C), humid (54% RH) environment. Subjects exercised on four occasions at weekly intervals with each trial beginning in the morning, 3 h after a standard breakfast. Over a 30-min period beginning 30 min after the end of exercise, subjects ingested one of the four test drinks in a volume equivalent to 1.5 times their body mass loss. Drink composition was constant except for the sodium (and matching anion) content. Sodium content of drinks A, B, C and D was 2, 26, 52 and 100 mmol.l-1, respectively. Treatment order was randomised using a four-way crossover incomplete block design. Blood and urine samples were obtained before exercise, immediately before and after the rehydration period and at 0.5, 1.5, 3.5 and 5.5 h after the end of the rehydration period. Data were analysed by parametric or non-parametric statistical tests are appropriate. The volume of fluid consumed was the same on all trials [2045(45) ml]. From the 1.5-h sample onwards, a significant treatment effect on cumulative urine output was apparent, with the volume excreted being inversely related to the sodium content of the drink consumed. By the end of the trial, subjects were in net negative fluid balance on trials A [by 689(124) ml] and B [by 359(87) ml]; on trials C [-2(79) ml] and D [+98(67) ml], subjects were approximately euhydrated. Cumulative urinary sodium output was higher on treatment D than on the other trials after 5.5 h. Plasma volume was lower after exercise than before; on trials B, C and D, plasma volume was higher than the pre-exercise value from 0.5 h after the end of the rehydration period onwards. On trial A, plasma volume was higher than the pre-exercise value at 3.5 and 5.5 h after the end of the rehydration period. At 1.5 h after the end of the rehydration period, the increase in plasma volume was greater on trials C and D than on trial A. These results suggest that the fraction of the ingested fluid that was retained was directly related to the sodium concentration.