GH production rates markedly increase during human puberty, mostly as an amplitude-modulated phenomenon. However, GH-deficient children have been dosed on a standard per kg BW basis similar to prepubertal children. This randomized study was designed to compare the efficacy and safety of standard recombinant human GH (rhGH) therapy (group I, 0.3 mg/kg x week) vs. high dose therapy (group II, 0.7 mg/kg x week) in GH-deficient adolescents previously treated with rhGH for at least 6 months. Ninety-seven children with documented evidence of GH deficiency (peak GH in response to stimuli, <10 ng/mL), with either organic or idiopathic pathology, were recruited. Both groups were matched for sex (group I, 42 males and 7 females; group II, 41 males and 7 females), age [group I, 14.0+/-1.6 (+/-SD) yr; group II, 13.7+/-1.6], standardized height (group I, -1.4+/-1.1; group II, -1.2+/-1.1), bone age (group I, 13.1+/-1.3 yr; group II, 13.1+/-1.3) etiology, maximum stimulated GH, previous growth rate, and midparental target height. All subjects were in puberty (Tanner stage 2-5) at study entry. Of the 97 subjects enrolled, 45 were treated for 3 yr or more; 48 completed the study. Of the subjects who discontinued the study, the most common reason was satisfaction with their height, although others discontinued for adverse events or personal reasons. The frequency of patients who discontinued was the same in both groups. The primary efficacy analysis was the difference between dose groups for near-adult height, defined as the height attained at a bone age of 16 yr or more in males and 14 yr or more in girls; all subjects who qualified were included in the analysis. This difference was statistically significant at 4.6 cm by analysis of covariance (ANCOVA; P < 0.001; n = 75). For subjects who received at least 4 yr of rhGH treatment, the difference between dose groups at that time point was 5.7 cm (by ANCOVA, P = 0.024; n = 20). The mean height SD score at near-adult height was -0.7+/-0.9 in the standard dose group and 0.0+/-1.2 in the high dose group. At 36 months the cumulative change in height (centimeters) was 21.5+/-5.3 cm (group I) vs. 25.1+/-4.9 (group II; P < 0.001, by ANCOVA); the change in Bayley-Pinneau predicted adult height was 4.8+/-4.2 cm (group I) vs. 8.4+/-5.7 (group II; P = 0.032). Median plasma IGF-I concentrations at baseline were 427 microg/L (range, 204-649) in group I and 435 microg/L (range, 104-837) in group II; at 36 months they were 651 microg/L (range, 139-1079) in group I vs. 910 microg/L (range, 251-1843) in group II (P = NS). No difference in change in bone age was detected between groups at any interval. High dose rhGH was well tolerated, with a similar safety profile as standard dose treatment and no difference in hemoglobin A1c or glucose concentrations between groups. In summary, compared to conventional treatment, high dose rhGH therapy in adolescents 1) increased near-adult height and height SD scores significantly, 2) did not increase the rate of skeletal maturation, and 3) appears to be well tolerated and safe. In conclusion, high dose rhGH therapy may have a beneficial effect in adolescent GH-deficient patients, particularly those who are most growth retarded at the start of puberty.