To clarify the relative usefulness of measuring stimulated as compared with spontaneous growth hormone levels in the diagnosis of growth hormone deficiency, we studied 54 short prepubertal children--23 with growth hormone deficiency identified by stimulation tests and 31 with idiopathic short stature who had normal responses to growth hormone stimulation. Growth hormone levels were measured in plasma samples obtained every 20 minutes for either 12 or 24 hours. The results were compared with those in 46 normal prepubertal children. Children with growth hormone deficiency had significantly lower mean 24-hour growth hormone levels (1.0 microgram per liter; range, 0.5 to 1.8) than normal children (2.8 micrograms per liter; range, 0.8 to 5.8; P less than 0.001). However, the diagnostic usefulness of the spontaneous growth hormone test was inferior to that of the stimulation tests, since it identified only 57 percent of the children with growth hormone deficiency identified by the stimulation tests. In the remaining children with growth hormone deficiency, spontaneous growth hormone levels were within the normal range. Children with idiopathic short stature had a normal mean 24-hour level of growth hormone (3.0 micrograms per liter; range, 1.1 to 6.7). No child in this group had low levels of spontaneous growth hormone secretion. We conclude that the measurement of the spontaneous secretion of growth hormone in prepubertal short children had lower sensitivity and offered no diagnostic advantage over stimulation tests. Our data do not support the routine measurement of spontaneous growth hormone secretion in the diagnosis of growth hormone deficiency.