There is no consensus as to the most appropriate method of diagnosing growth-hormone (GH) deficiency in adults. We have evaluated the relative diagnostic merits of measuring peak GH response to insulin-induced hypoglycaemia (insulin tolerance test), mean 24 h GH concentration derived from 20 min sampling, serum insulin-like growth factor I (IGF-I) concentrations, and serum IGF binding protein 3 (IGFBP-3) concentrations. These tests were undertaken in 23 patients considered GH deficient from extensive organic pituitary disease, and in 35 sex-matched normal subjects of similar age and body-mass index. Hypopituitary subjects had significantly lower stimulated peak GH, mean 24 h GH, IGF-I, and IGFBP-3 concentrations than normal subjects. The ranges of stimulated peak GH responses were clearly separated between the hypopituitary (< 0.2-3.1 ng/mL) and normal (5.3-42.5 ng/mL) groups, but mean 24 h GH, IGF-I, and IGFBP-3 concentrations overlapped. Mean 24 h GH concentrations were below assay sensitivity in 80% of hypopituitary subjects and 16% of normal subjects. 70% and 72%, respectively, of the IGF-I and IGFBP-3 values in hypopituitary subjects were within the range for normal subjects. We conclude that GH deficiency in adults is most reliably identified by stimulatory testing, and that IGF-I and IGFBP-3 are poor diagnostic tests of adult GH deficiency.