Recent reports, based on measurements of plasma GH levels, have challenged the concept that GH secretion is dependent on sleep and not modulated by circadian rythmicity. Because plasma levels reflect not only the secretory process, but also the effects of distribution and degradation, temporal limits of active secretion and, consequently, synchrony with other physiological events cannot be accurately estimated from circulating concentrations. The present study was undertaken to examine the roles of sleep and time of day in modulating pulsatile GH secretion, using a mathematical procedure (deconvolution) allowing secretory rates to be estimated from peripheral levels. Eight young nonobese healthy men participated each in six separate 16-h studies involving either normal or delayed sleep. Plasma GH levels were measured at 15-min intervals, and GH secretory rates were calculated by deconvolution. Each individual study was preceded by one night of habituation, and sleep was polygraphically recorded in all studies. Repeated measurements of plasma insulin-like growth factor-I (IGF-I) were performed in all subjects. Deconvolution revealed the existence of approximately 20% more GH pulses than detected in the plasma profiles. Large peaks of plasma GH concentrations often reflected the occurrence of a succession of secretory pulses. The total amount of GH secreted varied 10-fold across individual studies, but the within-subject variability (32%) was less than half the across-subject variability (65%). IGF-I levels were also more reproducible for a given subject than across subjects (11% vs. 36% variability) and did not correlate with the amount of GH secreted. During normal waking hours, the GH secretory rate was similar in the evening and the morning. This secretory rate was doubled during wakefulness at times of habitual sleep and tripled during sleep, even when sleep was delayed until 0400 h. A pulse starting within 30 min after sleep onset was present in all profiles with normal sleep and in 13 of 16 profiles with delayed sleep. The amount of GH secreted in response to sleep onset was tightly correlated with the level of secretion during wakefulness (r = 0.92). Almost 70% (57 of 83) of the pulses occurring during sleep were associated with slow wave (SW) stages. The amount of GH secreted in SW-associated pulses was correlated with the amount of SW occurring during the pulse, even when sleep-onset pulses were not considered. We conclude that in normal adult men, the amount of GH secretion and the levels of IGF-I are more reproducible within than across individuals.(ABSTRACT TRUNCATED AT 400 WORDS)