The purpose of this study was to define the relationship between arterial immunoreactive glucagon (IRG) and IRG that perfuses the liver via the portal vein during exercise in the diabetic state. Dogs underwent surgery >16 days before the experiment, at which time flow probes were implanted in the portal vein and the hepatic artery, and Silastic catheters were inserted in the carotid artery, portal vein, and hepatic vein for sampling. Dogs were made diabetic with alloxan injected intravenously approximately 3 wk before study (AD) or were studied in the nondiabetic state (ND). Each study consisted of a 30-min basal period and a 150-min moderate-exercise period on a treadmill. The findings from these studies indicate that the exercise-induced increment in portal vein IRG can be substantially greater in AD compared with ND, even when arterial and hepatic vein increments are not different. The larger IRG gradient from the portal vein to the systemic circulation in AD dogs is a function of a twofold greater increase in nonhepatic splanchnic IRG release and a fivefold greater hepatic fractional IRG extraction during exercise. In conclusion, during exercise, arterial IRG concentrations greatly underestimate the IRG levels to which the liver is exposed in ND, and this underestimation is considerably greater in dogs with poorly controlled diabetes.