Survival rates were compared in 82 patients who underwent therapeutic portacaval shunt. All patients were followed for at least 5 yr after shunt or until death. Survival rates were calculated by Life Table methods. Based on a combination of currently accepted histological and clinical criteria, there were 45 patients with Laennec''s cirrhosis; 29 patients with postnecrotic cirrhosis, 11 of whom had histological evidence of chronic active hepatitis; and 8 patients with primary biliary cirrhosis. Survival rates were similar in all 3 groups, alcoholic cirrhosis, postnecrotic cirrhosis and primary biliary cirrhosis. Hepatic reserve, as defined by Child''s classification, provided the best criteria for predicting survival. The type of shunt, end-to-side, side-to-side, or spleno-renal, did not influence survival. Histological evidence of chronic active hepatitis adversely affected survival in patients with postnecrotic cirrhosis. Histological evidence of ongoing alcoholic hepatitis in patients with Laennec''s cirrhosis did not influence survival adversely. Once a patient with cirrhosis has bled from esophageal varices, the etiology of the cirrhosis is apparently not a major factor in determining survival after a therapeutic portacaval shunt.