A total of 198 subjects were randomized to either high-dose (0.05 mg/kg.d) or low-dose (0.025 mg/kg.d) GH for 7 d; the alternate dose was then received after a 2-wk washout period. Groups included in the study were: normal, GH-insensitive (GHI; homozygous for the E180 splice mutation); heterozygous GHI (carriers of the E180 splice mutation); GH-deficient; and idiopathic short stature. Serum IGF binding protein-3 (IGFBP-3) concentrations (collected on d 1, 5, and 8 of treatment weeks) were GH-dependent, with significant elevation by d 5 of treatment, regardless of dose, in all normal subjects. GHI subjects had low baseline IGFBP-3 and poor or no response to either low- or high-dose GH. Heterozygous subjects, however, did not differ from age-matched normals with regard to IGFBP-3 generation. All GH-deficient subjects had subnormal baseline concentrations of IGFBP-3; most, but not all, were able to generate levels into normal ranges by 8 d of therapy. Children with idiopathic short stature showed a better response in IGFBP-3 generation compared with that previously observed with IGF-I, reaching concentrations in normal range with either dose of GH, suggesting that any GHI in this group is relatively limited to IGF-I production. For the diagnosis of GHI, the highest sensitivity (100%) and specificity (92%) was found on d 8 of the high-dose GH-IGFBP-3 generation test. Failure to raise both IGF-I and IGFBP-3 lowered sensitivity to 82-86% with low-dose GH, and 86-91% with high-dose GH.