The goal of our study was to compare the clinical usefulness of plasma insulin-like growth factor-I (IGF-I) (with and without binding protein extraction) and IGF binding protein-3 (IGFBP-3) measurements in the diagnosis of growth hormone (GH) disorders in adults. IGF-I and IGFBP-3 concentrations were measured in 25 acromegalic and 25 GH-deficient adult (GHDA) subjects (20-76 years) by comparison to a control population (n = 81) after age and sex stratification. In untreated acromegaly, IGF-I and IGFBP-3 were clearly increased (10 times the mean of controls for unextracted IGF-I, 4 times for extracted IGF-I and 2 times for IGFBP-3). Using the mean + 2SD of the control population as the cut-off point, the sensitivity of IGF-I for the diagnosis of acromegaly was higher than that of IGFBP-3 (unextracted IGF-I: 96% and extracted IGF-I: 100% vs IGFBP-3: 76%). In GHDAs, IGF-I and IGFBP-3 were decreased (34% of the mean of controls for unextracted IGF-I, 37% for extracted IGF-I and 70% for IGFBP-3). Using the mean - 2SD of the control population as the cut-off point, the sensitivity of IGF-I measurement for the diagnosis of GHDA was relatively low, but better for unextracted (68%) than for extracted IGF-I (52%). The sensitivity of IGFBP-3 was much lower (36%), thus invalidating this parameter for the diagnosis of GHDA. Our observations demonstrate that IGF-I measurement is a more powerful tool than IGFBP-3 measurement for the diagnosis of GH disorders in adults. Both IGF-I and IGFBP-3 are very useful for the diagnosis of acromegaly, but they are less reliable for diagnosing GHDA, as normal IGF-I or IGFBP-3 values do not rule out GH deficiency.