The plasma glucose and insulin responses to intravenous administration of tolbutamide in patients with insulinoma and healthy control subjects were compared to determine the sensitivity and specificity of the intravenous tolbutamide test for the diagnosis of insulinoma. The records of 406 healthy persons without concurrent disease known to affect glucose homeostasis or insulin concentrations and 41 patients with histologically confirmed insulinoma who underwent standard intravenous tolbutamide testing during the period from 1976 to 1986 were reviewed. The 5th percentile of the mean of plasma glucose levels at the 120-, 150-, and 180-minute points (G120-180) after injection of tolbutamide was 55 mg/dl for lean and 62 mg/dl for obese control subjects. With 95% specificity, the sensitivity of these criteria was 95% for lean and 100% for obese patients with insulinoma. Minimal differences were observed between men and women. At 95% specificity, the sensitivities were less for the ratio of the 180-minute plasma glucose to fasting plasma glucose level (64% in lean and 75% in obese patients), for mean of plasma insulin at the 120-, 150-, and 180-minute points (IRI120-180) (53% in lean and 36% in obese patients), for maximal insulin concentration (27% in lean and 31% in obese patients), for increase in insulin concentration above basal (18% in lean and 23% in obese patients), and for the combined criteria of the 5th percentile of G120-180 and the 95th percentile of IRI120-180 (47% in lean and 73% in obese patients).(ABSTRACT TRUNCATED AT 250 WORDS)