To evaluate the relative usefulness of insulin-like growth factor I (IGF-I) and IGF-binding protein-3 (IGFBP-3) in screening for GH status, GH stimulation (arginine-insulin/L-DOPA) tests and overnight GH studies (every 20 min sampling) were performed in 104 healthy short children (32 girls), aged 3-16 yr (height, -1.8 or more SD). IGFBP-3 had no advantage over IGF-I in screening sensitivity or specificity. IGF-I correlated with mean nighttime GH. Both IGF-I and IGFBP-3 correlated with peak stimulated GH. To identify more than 90% of children with GH deficiency (GHD) and borderline GHD, the mean values for age for IGF-I and IGFBP-3 were required as the cut-off criterion. However, at this criterion, 70% or more of idiopathic short stature (ISS) children would have to undergo testing to identify 90% of GHD or borderline GHD. More stringent criteria (-1.0, -1.64, and -2.0 SD) were more specific, but lost sensitivity. A practical application is suggested. Screening use of IGF-I with criterion of -1.0 SD would identify a subgroup that includes 88% of GHD, 71% of borderline GHD, and 46% of ISS. Both IGF-I and IGF-BP-3 higher than -1.0 SD would accurately identify 68% of ISS as not needing GH testing. Evaluation of growth velocity would identify the remaining children requiring definitive testing. Thus, combined screening for GHD using both IGF-I and IGFBP-3 has no better sensitivity than either test alone. However, such combined screening will improve the specificity and thus decrease the number of normal but short children who might otherwise undergo unnecessary testing.