The diagnosis of growth hormone (GH) deficiency in adults is based on provocative testing of GH secretion. The insulin tolerance test (ITT), currently the favoured test for this diagnosis, has been criticised for poor reproducibility and inconvenience. Since the combined administration of GH-releasing hormone (GHRH) plus GH-releasing peptide-6 (GHRP-6) is the most potent stimulus of GH secretion, we did a multicentre study comparing GH peaks elicited by ITT with those elicited by the GHRH/GHRP-6 test in healthy controls and GH-deficient individuals (cases). 125 adult patients with organic pituitary disease and 125 healthy individuals were studied. All cases and controls were given GHRH 1 microg per kg bodyweight intravenously plus GHRP-6 1 microg per kg intravenously at 0 min and blood samples were obtained during a subsequent 120 min period. 27 controls and all cases had an ITT. Inclusion criteria were severe GH deficiency--ie, a GH peak after ITT of < or = 3 microg/L. Results of the GHRH/GHRP-6 test were analysed by receiver-operating characteristic curve methodology. GH peaks seen after the GHRH/GHRP-6 test did not result in any side-effects and were not affected by age, sex, amount of adipose tissue, or by the GH assay system used. The GH mean peak after the GHRH/GHRP-6 test was 59.2 microg/L (SD 2.2) for controls and 4.1 microg/L (0.3) for cases, whereas after ITT the mean peak was 14.3 microg/L (1.7) and 0.5 microg/L (0.06), respectively. The differential peak responses of controls and cases was greater (p<0.001), for GHRH/GHRP-6 test than for ITT. When individually analysed GH peaks were a continuum, from 139.0 microg/L to 0.01 microg/L, with a cut-off point of 15.0 microg/L. The GHRH/GHRP-6 test performed well under the ROC curve analysis. For clinical utility, it is then proposed that values > or = 20.00 microg/L be considered normal and < or = 10.00 microg/L as GH deficient. The GHRH/GHRP-6 test is a convenient, safe and reliable test for adult GH deficiency and is not confounded by clinical factors known to alter GH secretory patterns. An evoked GH concentration of > or = 15.0 microg/L accurately distinguishes between healthy and GH-deficient adults.