The role of IGF-I and IGFBP-3 measurements in the diagnostic work-up of short children is established but remains controversial. Little information exists on the value of IGFBP-2 measurements. Based on reference data established in 388 children we have reinvestigated the issue, using data from 392 short children who underwent the same diagnostic procedures between 1987 and 1998 (GHD, n = 187; non-GHD, n = 205, including patients with ISS, n = 76; IUGR, n = 46; and TS, n = 83). In comparing IGF-I, IGFBP-3 and IGFBP-2 serum levels of GHD and ISS children with reference data, we calculated the sensitivity, specificity, efficiency and positive predictive value for the diagnosis of GHD. The overall sensitivity of the parameters was high, the rank order being as follows: IGF-I >IGFBP-3 >IGFBP-2 (75, 67 and 62%, respectively). In contrast, the specificity was relatively low: IGFBP-3 >IGFBP-2 >IGF-I (50, 50 and 32%, respectively). The efficiency and positive predictive value of parameters was in the order of 40, 60 and 70–80%, respectively. In repeated measurements, the recorded basal levels of IGF-I and IGFBP-3 showed an overall narrow range of variation. We conclude that the determination of basal IGF parameters is, together with anthropometry and imaging techniques, an indispensable tool for differentiating between GHD and ISS; and that IGFBP-2 plays an additional role in this process.