The isotope dilution technique has been extensively used to assess insulin action in humans. To determine if nonsteady state (NSS) has led to erroneous estimates of hepatic and extrahepatic insulin sensitivity, we measured glucose turnover in healthy subjects during infusion of insulin at rates of 0.25, 0.6, and 2.0 mU.kg-1.min-1. Turnover was calculated using Steele's traditional NSS equations [fixed-effective volume (pV) method] as well as with methods [radioactive infused glucose (hot-GINF) or variable pV] designed to minimize NSS error. In contrast to the fixed-pV method, both the hot-GINF and variable-pV methods indicated that several hours were required for suppression of hepatic glucose release at all insulin concentrations and that small increases in plasma insulin (approximately 100 pmol/l) had comparable effects on glucose disappearance and hepatic glucose release. Nevertheless, despite these differences, when turnover during the final hour of the insulin infusions was plotted vs. the prevailing insulin concentration, all three methods yielded similar insulin dose-response curves for suppression of hepatic glucose release. Thus despite previous errors in measurement of glucose turnover, the widely accepted belief that the human liver is exquisitely sensitive to small changes in insulin is correct.