We studied effects of changes in hepatic insulin influx on hepatic insulin retention. In order to compare insulin net flux (influx - efflux) with percentage (%) hepatic insulin extraction (net flux/influx X 100), as indices of hepatic insulin retention, we examined which index is better predicted by insulin influx to the liver. Electromagnetic hepatic blood flow measurements were made and blood was sampled from portal, hepatic and peripheral veins and femoral artery in anesthetized dogs. Insulin influx was perturbed by sequentially superimposed peripheral venous infusions of somatostatin, glucagon and insulin (which were subsequently discontinued one at a time). Although the somatostatin was biologically effective in inhibiting insulin and glucagon release we found no significant plasma flow changes on starting or stopping the infusion of somatostatin (800 ng X kg-1 X min-1). Net insulin flux was linearly related to influx (n = 56, r = 0.99, p less than 0.001). Hepatic glucose production was better correlated with net insulin flux (r = -0.66; p less than 0.01) than with % extraction (r = 0.08; N.S.), when mean glucagon influx was held constant (partial correlation). Although net insulin flux and % extraction are complementary indices, the former reflects influx changes more faithfully than the latter.