Five subjects performed intermittent exercise on a bicycle ergometer (25 min work, 5 min rest cycles for 2 hours, and 20 min work, 10 min rest cycles for a further hour) in a hot environment (air and wall temperatures = 36 degrees C; dew-point temperature = 10 degrees C; air velocity = 0.6 m.s-1). The relative mechanical work load was of 70 W (30% of the maximal aerobic capacity). Seven experimental tests were carried out in order to induce a plasma hypovolemia associated with either a plasma hypo- or hyperosmolarity. The preexercise level of body hydration was also manipulated by giving a diuretic, or by ingestion of 500 ml of isotonic electrolyte sucrose solution before the start of exercise. Continuous measurements were made of rectal and mean skin temperatures. The sweating responses of the chest and of the thigh (over the active muscles of the leg) were monitored from 4 sweat collection capsules highly ventilated. On each of these body areas, the local skin temperatures under one of the 2 capsules was kept at a constant level (37 degrees C). The effects of the level of body hydration on the sweating response only appear when a high local thermal clamp is imposed beneath the capsule. This local effect is particularly strong over the active muscles of the thigh. The influence of the preexercise hydration appears during dehydration tests. This effect is not significant when fluid is given to the subject during the exercise. The change in the sensitivity of the thermoregulatory system is more strongly associated with plasma osmolarity than hypovolemia.(ABSTRACT TRUNCATED AT 250 WORDS)