Short-term administration of human growth hormone to children with idiopathic short stature can improve mean growth rate and predicted adult height. It is yet unknown whether therapy would alter pubertal development or affect final height. Three-year treatment results in a group of children with idiopathic short stature are reported. For year 1 of the study, 121 prepubertal children were randomly selected to receive somatotropin, 0.3 mg/kg per week, administered subcutaneously three times weekly (n = 63), or to be nontreatment control subjects (n = 58). After 1 year, all subjects were again randomly selected to receive either three-times-weekly or daily dosing at the same total dose. For the 92 subjects who completed 36 months of treatment, mean growth rate increased from a mean of 4.6 cm/yr before treatment to a mean of 8.0 cm/yr in the first year of treatment. Daily dosing resulted in a significantly faster mean growth rate (9.0 cm/yr) than three-times-weekly dosing (7.8 cm/yr) (p = 0.0005). Mean growth rates were 7.6 and 7.2 cm/yr during years 2 and 3, respectively, and did not differ by dosing group. Mean standardized height for all subjects improved from -2.7 to -1.6 after 3 years. When the growth rate was standardized for bone age, however, subjects who remained prepubertal had a significantly greater gain in mean height SD score than subjects who became pubertal during that 3-year period (p < 0.02). Mean standardized Bayley-Pinneau predicted adult height SD score increased from -2.7 to -1.6 and was independent of the timing of pubertal onset, but for individuals this score was more variable. Year-1 growth response, expressed as growth rate or change in height SD score, was the best predictor of growth in subsequent years. Responses to therapy could not be reliably predicted from baseline anthropometric variables, plasma insulin-like growth factor I SD score, growth hormone levels. Final height assessment will be needed to determine the ultimate benefit of therapy.