Thirteen boys with constitutional delay in growth and/or development aged 7.6 to 16 years received 2.5 mg/day of oxandrolone for 3 months. Their growth response on treatment and in the subsequent 3 months was compared to that of 14 age-matched untreated controls. Growth rates were analysed in groups arbitrarily defined by testicular volume at entry (less than 4 ml prepubertal, greater than 4 ml pubertal). Growth velocities increased on oxandrolone (prepubertal, mean 4.4 to 7.5 cm/year, P = 0.05: pubertal, mean 4.7 to 8 cm/year, P less than 0.05). Over the next 3 months, the pubertal boys grew at 9.3 cm/year, while the prepubertal group decelerated to 6.2 cm/year. Both control groups showed no significant change in velocity over 6 months. GH responses to arginine and GRF, and to sleep in prepubertal boys, were unchanged throughout the study. However, in pubertal boys the mean GH levels ('area under the curve') during sleep at 3 and 6 months, had increased over baseline values, associated with a significant increase at 6 months in basal somatomedin-C (140 to 214 ng/ml, P less than 0.05). Oxandrolone does not alter the GH status of prepubertal boys, and thus probably promotes growth by a direct action at the growth plate. In contrast, the persistent growth acceleration of pubertal boys may be associated with increased GH and somatomedin-C levels: in this group, oxandrolone has proved a useful stimulus to growth.