OBJECTIVE: The insulin-tolerance test (ITT) is currently considered to be the gold standard for evaluating adults suspected of GH deficiency (GHD). The aim of this study was to determine factors that may influence nadir blood glucose (BG) when using a mean insulin dose of 0.1 IU/kg body weight. Furthermore, we wanted to evaluate the safety and GH-related aspects of the ITT. DESIGN: ITT was performed in 277 patients, of whom 255 (129 females) were eligible for evaluation. RESULTS: Multiple regression analysis, including the whole population, showed that the major determining factors for nadir BG were basal BG and body mass index (BMI) (P<0.02). No serious adverse event was recorded. Sixty-three percent of all patients tested had severe GHD with peak GH response to hypoglycaemia below 7.8 mIU/l. The positive predictive value for IGF-I was 0.82 and the negative predictive value was 0.47, using a cut-off value corresponding to -2 s.d. GH peak response to hypoglycaemia decreased with increasing numbers of other pituitary hormone deficiencies. CONCLUSIONS: When determining the dose of insulin based on weight, factors like pre-test BG and BMI should also be considered. We propose an algorithm stating that the dose of insulin should be 0.1 IU insulin/kg body weight minus 2 IU if pre-test BG is <4.0 mmol/l and minus 2 IU if BMI is <20 kg/m(2) in order to take these factors into account. Our findings furthermore support the concept that the low-dose ITT is a safe test in adults, when performed in experienced hands. It was confirmed that IGF-I is not sufficient when diagnosing GHD in adults, and reliable stimulation tests like ITT are required in the diagnosis.