Within an appropriate clinical context, severe GH deficiency (GHD) in adults has to be defined biochemically by provocative testing of GH secretion. Patients with childhood-onset GHD need retesting in late adolescence or young adulthood to verify whether they have to continue recombinant human GH treatment. GHRH + arginine (GHRH+ARG) is the most reliable alternative to the insulin-induced hypoglycemia test (ITT) as a provocative test for the diagnosis of GHD in adulthood, provided that appropriate cut-off limits are assumed (normal limits, 16.5 microg/L as 3rd and 9.0 microg/L as 1st centile). We studied the GH response to a single GHRH (1 microg/kg iv) + ARG (0.5 g/kg iv) test in 62 young patients who had undergone GH replacement in childhood, based on the following diagnosis: 1) organic hypopituitarism with GHD (oGHD) In = 18: 15 male (M), 3 female (F); age, 26.8+/-2.2 yr; GH peak < 10 microg/L after two classical tests]; 2) idiopathic isolated GHD (iGHD) [n = 23 (15 M, 8 F); age, 23.0+/-1.5 yr; GH peak < 10 microg/L after two classical tests]; and 3) GH neurosecretory dysfunction (GHNSD) [n = 21 (10 M, 11 F); age, 25.1+/-1.6 yr; GH peak > 10 microg/L after classical test but mGHc < 3 microg/L]. The GH responses to GHRH+ARG in these groups were also compared with that recorded in a group of age-matched normal subjects (NS) [n = 48 (20 M, 28 F); age, 27.7+/-0.8 yr]. Insulin-like growth factor I levels in oGHD subjects (61.5+/-13.7 microg/L) were lower (P < 0.001) than those in iGHD subjects (117.2+/-13.1 microg/L); the latter were lower than those in GHNSD subjects (210.2+/-12.9 microg/L), which, in turn, were similar to those in NS (220.9+/-7.1 microg/L). The mean GH peak after GHRH+ARG in oGHD (2.8+/-0.8 microg/L) was lower (P < 0.001) than that in iGHD (18.6+/-4.7 microg/L), which, in turn, was clearly lower (P < 0.001) than that in GHNSD (31.3+/-1.6 microg/L). The GH response in GHNSD was lower than that in NS (65.9+/-5.5 microg/L), but this difference did not attain statistical significance. With respect to the 3rd centile limit of GH response in young adults (i.e. 16.5 microg/L), retesting confirmed GHD in all oGHD, in 65.2% of iGHD, and in none of the GHNSD subjects. With respect to the 1st centile limit of GH response (i.e. 9.0 microg/L), retesting demonstrated severe GHD in 94% oGHD and in 52.1% of iGHD. All oGHD and iGHD with GH peak after GHRH+ARG lower than 9 microg/L had also GH peak lower than 3 microg/L after ITT. In the patients in whom GHD was confirmed by retesting, the mean GH peak after GHRH+ARG was higher than that after ITT (3.4+/-0.5 vs. 1.9+/-0.4). In conclusion, given appropriate cut-off limits, GHRH+ARG is as reliable as ITT for retesting patients who had undergone GH treatment in childhood. Among these patients, severe GHD in adulthood is generally confirmed in oGHD, is frequent in iGHD, but never occurs in GHNSD.