From 1961 to 1971, 73 central splenorenal and 66 portacaval shunts were done for cirrhotic patients who had bled from esophageal varices. Comparative analysis revealed low (3% and 2%, respectively) operative mortality for elective operations, equal (93%) effectiveness in control of variceal bleeding, but substantial diferences in the incidence of postshunt encephalopathy. For patients who had mimal hepatic dysfunction before operation, disabling disorders in mentation developed in 5% of patients who had splenorenal shunts, in contrast to 50% of patients who had portacaval shunts. Survival rates after the two shunts were nearly identical. Thus, the advantages of splenorenal shunts concern the quality of life but not the length of survival. These observations are considered in relation to available therapeutic alternatives.