To estimate the incidence of low growth hormone (GH) concentration in children and adolescents with idiopathic short stature, overnight GH levels were measured in 167 subjects. The results were compared with data from 132 normal children of similar pubertal stage, bone age, or body mass index. The majority of short children had normal overnight GH concentrations in a distribution not significantly different from that observed in normal children. However, in 6% of children grouped by pubertal stage and in 13% of children grouped by bone age, overnight GH levels were below the 95% confidence limits of normal. The overnight GH levels were above normal in 6%. The observed frequencies of both low and high GH levels were significantly greater than expected (P < 0.001). However, when body mass index was included in the analysis, only 5% of the children had low GH measures, and only 4% had high GH measures (both not significant). This frequency of low overnight GH levels in short children is considerably less than that reported by others. Thus, these data do not support the hypothesis that a deficiency of spontaneous GH secretion is a common cause of short stature. We conclude that standard GH stimulation tests, despite their limitations, remain the best definitive test of GH secretion. Subsequent overnight GH studies may be useful, however, in selected clinical settings such as previous cranial irradiation or other central nervous system disorder.