GH spontaneous peaks and their diagnostic utility have not been previously evaluated by means of a 24-h continuous withdrawal (CW) procedure in children with growth disorders. Using a CW pump, we studied the 24-h spontaneous secretion of GH in 129 prepubertal subjects grouped as follows. The control group (C) consisted of 20 children of normal height and growth rate. Group GHD consisted of 53 patients with classical GH deficiency (48 idiopathic and 5 organic). The NSD group consisted of 36 patients with a growth velocity below 4.5 cm/yr, normal GH response to provocative stimuli (PS), but a mean 24-h GH in the deficient range. Group NSS consisted of 20 short children with normal growth velocity, normal PS, and normal mean 24-h GH concentration. The mean GH levels for the 24-h period were 4.1 +/- 1.7, 1.4 +/- 0.5, 2.1 +/- 0.7, and 4.2 +/- 1.9, respectively, for the C, GHD, NSD, and NSS groups. For each subject, GH levels were determined in 48 0.5-h samples collected during the CW study, and the GH profile was analyzed by the Pulsar computer program. The mean number of peaks was 9.0 +/- 2.5 for C, 9.5 +/- 3.7 for GHD, 10.5 +/- 1.8 for NSD, and 9.5 +/- 3.2 for NSS. There was no statistical difference between groups. The mean amplitude of peaks was 9.8 +/- 8.9 for C, 1.6 +/- 1.0 for GHD, 2.9 +/- 1.3 for NSD, and 9.9 +/- 9.1 for NSS. Mean peak amplitudes in both GHD and NSD were significantly lower than in the C and NSS groups. The presence of peaks of more than 8 micrograms/L during the daytime (0800-2000 h) was a characteristic of children with normal integrated GH concentration (IC-GH) and was seen in 90% of normally growing children but in only 7% of poorly growing children with subnormal IC-GH. There was no significant difference in the number of pulses during the night between C and NSD groups. We conclude that differences in IC-GH between normally growing and poorly growing children are due to a lower amplitude of peaks during the daytime hours.