A portable constant withdrawal pump was used to determine consecutive 30-min integrated concentrations of growth hormone (ICGH) for a 24-h period in 18 normal subjects, ages 5 to 28 yr. Seven prepubertal, 5 pubertal and 6 young adult subjects were studied under conditions of normal activity. Pubertal, prepubertal, and adult subjects had peaks during waking hours as well as during sleep, but peaks in the adults were usually lower than in the children. Using an analysis of variance, a significant (P smaller than 0.01) downward trend of ICGH was observed to occur during sleep. During waking hours a significant variation (P smaller than 0.01 by analysis of variance) was found with low levels in the first few hours after awakening and an upward trend as the evening approached. Food intake had no significant effect on ICGH nor did the introduction of the indwelling catheter. The mean apparent half-life values of growth hormone calculated on the basis of 87 episodes with 3 or more points on the downslope was 40.1 min. This is significantly higher than the known true half-life of the hormone, suggesting that complete secretory inactivity after a secretory episode is an infrequent event.