We describe a new noninvasive method for comparing insulin secreted acutely by the pancreas vs. a matched peripheral venous insulin infusion in humans. An intravenous tolbutamide infusion algorithm was developed that produced sustained steady rates of portal insulin secretion over 5 h in 11 healthy young men. Plasma glucose levels were maintained in the euglycemic range by adjusting the rate of an iv dextrose (20%) infusion. The pancreatic insulin secretory rate was calculated from peripheral C-peptide levels by deconvolution using standard parameters for a two-compartment mathematical model for C-peptide distribution and metabolism. On a subsequent occasion in the same subject, exogenous insulin was infused peripherally at a rate that matched the earlier tolbutamide-induced pancreatic insulin secretory rate, and euglycemia was maintained with a variable 20% dextrose infusion. The assumption that tolbutamide, when used in this fashion, has no independent insulin-like or insulin-potentiating effect at either low or high levels of peripheral insulinemia, does not affect insulin clearance, and does not suppress peripheral glucagon levels was validated in four patients with insulin-dependent diabetes mellitus. Mean peripheral immunoreactive insulin was significantly higher in the insulin infusion study than in the tolbutamide study (286 +/- 31 vs. 156 +/- 21 pmol/L; P = 0.0001). The dextrose infusion rates required to maintain euglycemia were also higher in the insulin infusion study (0.44 +/- 0.03 vs. 0.32 +/- 0.03 mmol/kg.min; P = 0.003). The MCR of insulin was greater in the tolbutamide infusion vs. the exogenous insulin infusion study (32.6 +/- 2.9 vs. 19.8 +/- 2.2 mL/kg.min; P = 0.0003), probably due to the hepatic first pass effect on insulin clearance when insulin is delivered by the portal route. This noninvasive method can be used in future studies to examine the relative importance of direct hepatic vs. peripheral effects of insulin in controlling hepatic glucose and lipid production.