The present study investigated the relationship between the milk protein content of a rehydration solution and fluid balance after exercise-induced dehydration. On three occasions, eight healthy males were dehydrated to an identical degree of body mass loss (BML, approximately 1·8%) by intermittent cycling in the heat, rehydrating with 150% of their BML over 1 h with either a 60 g/l carbohydrate solution (C), a 40 g/l carbohydrate, 20 g/l milk protein solution (CP20) or a 20 g/l carbohydrate, 40 g/l milk protein solution (CP40). Urine samples were collected pre-exercise, post-exercise, post-rehydration and for a further 4 h. Subjects produced less urine after ingesting the CP20 or CP40 drink compared with the C drink (P<0·01), and at the end of the study, more of the CP20 (59 (SD 12)%) and CP40 (64 (SD 6)%) drinks had been retained compared with the C drink (46 (SD 9)%) (P<0·01). At the end of the study, whole-body net fluid balance was more negative for trial C (- 470 (SD 154) ml) compared with both trials CP20 (- 181 (SD 280) ml) and CP40 (2107 (SD 126) ml) (P<0·01). At 2 and 3 h after drink ingestion, urine osmolality was greater for trials CP20 and CP40 compared with trial C (P<0·05). The present study further demonstrates that after exercise-induced dehydration, a carbohydrate--milk protein solution is better retained than a carbohydrate solution. The results also suggest that high concentrations of milk protein are not more beneficial in terms of fluid retention than low concentrations of milk protein following exercise-induced dehydration.