Seventy-eight prepubertal, non-GH-deficient children aged 8.1 +/- 0.2 y, with very short stature (mean, -3.2 SD) of intrauterine onset, were treated for 3 y with GH [0.4 (dose D1) or 1.2 (dose D2) IU/kg/wk] and 66 were followed during a 4th y without GH therapy. A 2-y intermediary report had demonstrated a GH dose-dependent acceleration of growth. During the 3rd y on GH, patients D2 (1.2 IU/kg/wk) continued with the same dose, whereas patients D1 (0.4 IU/kg/wk) were randomized to either continue on D1 (group D1) or be increased to D2 (group D1D2). After 3 y on GH, patients' mean height (SD) reached -2.37 (D1), -2.17 (D1D2), and -1.58 (D2) with a total mean height gain of 0.77 (D1), 0.93 (D1D2) (difference NS), and 1.61 SD (D2 significantly higher than D1 and D1D2, p < or = 0.0001). During the off-treatment year, mean growth rate (cm/y) decreased to 3.4 in patients D1, 3.7 in D1D2, and 4.1 in D2 (NS). During the 4 y, bone age advanced of 4.6, 4.6, and 5.3 y in D1, D1D2, and D2, respectively, and puberty started in 34 patients (10 during the off-treatment year). Age at onset of puberty, apparently within normal range, did not relate either to the dose or the duration of treatment. Clinical and biologic tolerance of treatment was good. In conclusion this study demonstrates a GH dose-dependent effect on growth acceleration in persistent postnatal severe growth retardation of intrauterine onset. This effect was sustained for 3 y at 1.2 IU/kg/wk followed by a peculiar growth deceleration at treatment discontinuation. Additional studies are necessary to optimize long-term GH treatment regimen and to document its effects on final height.