The authors investigated insulin secretory capacity and insulin action in 11 preoperative patients with pancreatic carcinoma and 15 age-matched and weight-matched healthy subjects (C). Five patients were classified as diabetic (D), two as impaired glucose tolerant (IGT), and four as nondiabetic (ND). Postabsorptive serum insulin levels (mean +/- SE, in uU/ml) in D (12 +/- 2), IGT (17 +/- 7), and ND (10 +/- 2) were comparable. After administration of 100 g of oral glucose, peak insulin achieved in D (60 +/- 11) was lower than in IGT (101 +/- 26) and ND (83 +/- 20), whereas peak insulin levels in IGT and ND were significantly (P less than 0.05) higher than in C (45 +/- 6). Comparable insulin response to nonglucose stimuli was documented in all subjects using the slow arginine infusion test with mean serum insulin of 27 +/- 4 in D, 28 +/- 6 in IGT, 34 +/- 10 in ND, and 32 +/- 5 in C. In six patients (P) and six controls, insulin action was assessed by the euglycemic hyperinsulinemic clamp technique, with glucose turnover rates estimated by [3-3H]glucose infusion. Steady-state plasma glucose concentrations were maintained at 92 +/- 3 (P) and 91 +/- 1 mg/dl (C). After insulin infusion at the rate of 1.0 mU/kg/min, comparable high physiologic insulin levels were observed in P (73 to 104 uU/ml) and in C (81 to 103 uU/ml). Postabsorptive rates of endogenous glucose appearance (Ra) were higher in P (2.86 to 3.02 mg/kg/min) than in C (1.50 to 2.80 mg/kg/min). At high physiologic insulin concentrations, negative Ra values were documented in all subjects, and complete suppression of Ra was assumed. Total body glucose use (M) was consistently lower in P (3.90 to 6.40 mg/kg/min) than in C (6.98 to 10.40 mg/kg/min), consistent with a state of insulin resistance. Patients with pancreatic cancer manifest insulin resistance by virtue of a decrease in total body glucose use (M) and decreased insulin response to glucose due to either inherent beta cell dysfunction or decreased islet cell mass. The latter is not identifiable by histologic morphology.