There is little information concerning the physiological response to hypoglycaemia induced by sulphonylureas. We compared the physiological and symptomatic responses to insulin and tolbutamide induced hypoglycaemia in 8 normal subjects. While infusing either insulin or tolbutamide, we used a glucose clamp to maintain blood glucose at 4.5 mmol l-1 for 30 min and lowered it to 2.9 mmol l-1 for a further 30 min. Mean peripheral insulin levels during the insulin infusion arm in comparison with the tolbutamide infusion were not significantly different during the euglycaemic plateau: 106 +/- 4 vs 77 +/- 15 mU l-1 (mean +/- SEM) (mean difference 29 mU l-1, 95% CI -22 to 80; p = NS) but were greater during the hypoglycaemic plateau: 106 +/- 3.5 vs 21.0 +/- 4.0 mU l-1 (mean difference 85 mU l-1, 95% CI 72 to 98; p < 0.0001). Portal insulin concentrations, calculated from C-peptide data were not significantly different during the euglycaemic plateau with insulin as compared to tolbutamide. However, during hypoglycaemia portal insulin concentrations were significantly higher 15 min from the start of the plateau, during insulin infusion. During hypoglycaemia induced by either insulin or tolbutamide there were similar peak responses of glucagon: 124 +/- 14 vs 128 +/- 7 ng l-1 (mean difference -4, 95% CI -39 to 31; p = NS) and adrenaline: 2.9 +/- 0.4 vs 2.8 +/- 0.3 nmol l-1, (mean difference 0.1, 95% CI -0.9 to 1.0; p = NS). Increases in tremor and sweating and deterioration in reaction time were similar during both periods of hypoglycaemia as were increases in total: 18.5 +/- 1.4 vs 19.6 +/- 2.2 (mean difference -1.0, 95% CI -3.8 to 1.8; p = NS) and autonomic: 8.9 +/- 0.9 vs. 9.9 +/- 1.3 (mean difference -1.1, 95% CI -5.9 to 3.6; p = NS) symptom scores. We conclude that there is no difference in the glucagon, sympathoadrenal, cognitive or symptomatic response during hypoglycaemia induced by either insulin or tolbutamide. This suggests that the different insulin concentrations produced by these contrasting models of hypoglycaemia had no effect on the physiological response and patients taking sulphonylureas can be expected to develop similar warning symptoms to those on insulin.