The 24-h rhythm of growth hormone (GH) is thought to be controlled primarily by sleep processes with a weak circadian component. This concept has been recently questioned in sleep-deprived persons. To test the notion of a high sleep-dependency of GH release, we established simultaneous 24-h rhythms of GH and melatonin, a circadian marker, in night workers who form a model for challenging sleep and circadian processes. Ten day-active subjects and 11 night workers were studied during their usual sleep-wake schedule, with sleep from 23:00 to 07:00 hours and 07:00 to 15:00 hours, respectively. Experiments were conducted in sleep rooms under continuous nutrition, bed rest, and dim light. Melatonin and GH were measured every 10 min over 24 h. In day-active subjects, melatonin and GH showed the well-known 24-h profiles, with a major sleep-related GH pulse accounting for 52.8 +/- 3.5% of the 24-h GH production and the onset of the melatonin surge occurring at 21:53 hours +/- 18 min. In night workers, melatonin showed variable circadian adaptation, with the onset of secretion varying between 21:45 and 05:05 hours. The sleep-related GH pulse was lowered, but the reduction was compensated for by the emergence of large individual pulses occurring unpredictably during waking periods, so that the total amount of GH secreted during the 24 h was constant. One cannot predict the degree of GH adaptation from the highly variable melatonin shift. These results argue against the concept that sleep processes exert a predominant influence on GH release whatever the conditions. When sleep and circadian processes are misaligned, the blunting of the sleep-related GH pulse is counteracted, as in sleep-deprived persons, by a compensatory mechanism promoting GH pulses during wakefulness.