Based on previous consensus statements, it has been widely accepted that the diagnosis of adult growth hormone deficiency (GHD) must be shown biochemically by provocative tests of GH secretion; in fact, the measurement of IGF-I as well as of other markers was considered unable to distinguish between normal and GHD subjects. The Insulin Tolerance Test (ITT) was indicated as that of choice and severe GHD defined by a GH peak lower than 3 microg/l. It is now recognized that, although normal IGF-I levels do not rule out severe GHD, very low IGF-I levels in patients highly suspected for GHD (i.e. patients with childhood-onset severe GHD or with multiple hypopituitarism acquired in adulthood) can be considered as definite evidence for severe GHD. However, patients suspected for adult GHD with normal IGF-I levels must be investigated by provocative tests. ITT remains a test of reference but it should be recognized that other tests are as reliable as ITT. Glucagon as classical test and, particularly, new maximal tests such as GHRH in combination with arginine or GH secretagogues (GHS) (i.e. GHRP-6) have well defined cut-off limits, are reproducible, able to distinguish between normal and GHD subjects. Overweight and obesity have confounding effect on the interpretation of the GH response to provocative tests. In adults cut-off levels of GH response below which severe GHD is demonstrated must be appropriate to lean, overweight and obese subjects to avoid false positive diagnosis in obese adults and false negative diagnosis in lean GHD patients.