We simultaneously measured hepatic insulin removal invasively and estimated hepatic clearance and extraction of insulin pharmacokinetically from cardiac output and peripheral plasma concentrations (relatively) noninvasively. The invasive methods involved continuous electromagnetic measurements of portal venous and hepatic arterial blood flow and simultaneous intermittent sampling of blood from the portal and hepatic veins and femoral artery for assay of insulin concentrations. The noninvasive method assumed that hepatic plasma flow is proportional to cardiac output and that hepatic clearance is a constant fraction of total body clearance of insulin. In anesthetized dogs (n = 6), endogenous insulin was suppressed with somatostatin (800 ng/kg/min) while biosynthetic human insulin (0.25, 0.50, and 1.00 mU/kg/min) was infused to steady state during three consecutive 90-min periods. Insulin concentrations were directly proportional to the infusion rate (p less than 0.01). Hepatic blood flow accounted for 20 +/- 2% of cardiac output. Measured hepatic clearance accounted for 51 +/- 5% of total body clearance of insulin and correlated with the pharmacokinetic estimates (p less than 0.01); the estimates of hepatic clearance ranged from 91 to 114% of the measured values. We conclude that this pharmacokinetic approach, which requires only samples of peripheral blood and estimates of hepatic blood flow, may be used to study the hepatic removal of insulin relatively noninvasively.