The aims of this study were to describe the basal serum levels of insulin-like growth factor I (IGF-I) and IGF-binding protein-3 (IGFBP-3) and to evaluate their changes during daily GH treatment (0.1 IU/kg; 33 micrograms/kg) for up to 2 yr in prepubertal short children born small for gestational age (SGA) and to correlate these changes with the growth response to GH therapy. Seventy-two prepubertal short children (height SD score, -5.4 to -2.0; age, 2.0-12.9 yr) born SGA (54 boys and 18 girls), eight of whom (six boys and two girls) had signs of Silver-Russell syndrome, participated in the study. The serum concentrations of IGF-I and IGFBP-3 were converted into SDS using our reference values for prepubertal healthy children. The mean (+/-SD) change in height SDS during the year before the start of GH treatment was 0.1 (0.2) and increased to 0.8 (0.4) during the first year (P < 0.001) and to 0.6 (0.3) during the second year of therapy (P < 0.001). Basal levels of both IGF-I and IGFBP-3 were significantly reduced compared with the reference values. The mean (+/-SD) basal serum concentration of IGF-I was -0.6 (1.1) SD score, and 80% of the SGA children had IGF-I levels below the 50th percentile of the reference group. The corresponding values for IGFBP-3 were -0.4 (1.0) SD score and 63%. The mean IGF-I level increased significantly by 55% from baseline to day 10 of treatment, by 76% on day 90, by 90% after 1 yr, and by 123% after 2 yr of GH treatment. The mean increases in IGFBP-3 were not as great: 25%, 27%, 35%, and 43%, respectively. The 1-yr growth response, expressed as the change in height SD score, correlated negatively with both the basal serum concentration of IGF-I (r = -0.37; P < 0.001) and IGFBP-3 (r = -0.35; P < 0.01), whereas a positive correlation was found to the 10-day percent increase in IGF-I (r = 0.32; P < 0.05). No correlations were found with the initial changes in IGFBP-3. Using a multiple stepwise linear regression analysis, the model using chronological age at the start of GH therapy, the mother's height expressed as a SD score, and the short term percent increase in IGF-I accounted for 42% of the variance in the 1-yr growth response. With the inclusion of 24-h GH profiles, 59% of the variability of the growth response could be explained. It is concluded that short prepubertal children born SGA show an increased growth rate in response to GH therapy. Their mean IGF-I and IGFBP-3 levels before treatment were low and correlated negatively with the growth response to treatment, indicating GH insufficiency. Finally, up to 59% of the variability in the 1-yr growth response to GH treatment could be explained by models using auxological and biochemical variables.