The diagnosis of GH deficiency (GHD) in children is based on clinical findings, auxology, hormonal results, brain magnetic resonance imaging (MRI), molecular analysis, and two separate GH provocative tests, for which a GH cut-off level of less than 10 ng/ml is the gold standard for defining GHD in children. However, great variability exists in the GH response to stimulation, due to provocative test type and variability of GH peak with age, sex, body composition and pubertal stage. Variability in GH measurements is attributed to GH assay type, GH molecule heterogeneity, GH-binding protein interference, GH calibrators, antibody recognition of different GH isoforms, and inconsistency of unit-conversion factors. Each parameter has a significant effect on GH measurement and consequently, on the diagnosis of GHD and the decision for GH therapy. Most routine laboratories have replaced the manual radioimmunoassay (RIA) with fully automated two-site sandwich immunoassays: the two are well-correlated but the latter provide significantly lower results than the former, which has enormous implications for interpretation of the GH provocative tests, the diagnosis of GHD, and the decision for GH therapy. Despite these lower values, the cut-off level for GHD diagnosis has not been revisited. Here, the reasons for the variability in GH measurements and the need to revisit GH cut-off values are discussed.