Controversial data exist in the literature about the presence and clinical relevance of hepatic arterial-venous shunting. An interesting opportunity for reconsidering the problem has been provided by the use, in the study of liver function, of D-sorbitol, a substance whose first-pass hepatic extraction is very high in normal subjects, while being directly related to circulatory alterations in liver cirrhosis. Because of this property, the systemic bioavailability of D-sorbitol during hepatic arterial infusion can be assumed to reflect arterial-venous shunting. Thirteen biopsy-proven cirrhotic patients (ages 35-66 years), who required diagnostic arterial catheterization, entered the study. Patients were studied on two subsequent days, in which a sterile pyrogen-free solution (1.5%) of D-sorbitol was administered by direct low-rate infusion (15 mg/min for 20 min) into the hepatic artery and the systemic circulation, respectively. Urine samples were spontaneously collected for 8-hr periods before and during/after each infusion. The hepatic arterial bioavailability of D-sorbitol was calculated as the ratio between the net cumulative urinary outputs of D-sorbitol after infusions into the hepatic artery and the systemic vein. Observed values confirm the existence and the large variability (0-88.7%) of hepatic arterial-venous shunting in cirrhotic patients.