Background: The aim of this study was to determine, in a prospective randomized clinical trial, whether the partial portacaval shunt offers any advantage in terms of liver function and encephalopathy rate when compared with direct side-to-side direct portacaval shunt. Methods: Forty-six “good risk” patients with cirrhosis and with documented variceal hemorrhage were randomly assigned to either a partial shunt procedure (achieved by 10-mm diameter interposition portacaval H-graft) or direct small-diameter side-to-side portacaval anastomosis. Results: Operative mortality was zero in both groups. During the follow-up period, encephalopathy developed in 3 patients in the partial shunt group and 9 in the direct shunt group (P =.04). Kaplan-Meier analysis demonstrated that encephalopathy-free survival was significantly longer in the partial shunt group (P =.025). Direct shunt patients had significant hepatic functional deterioration postoperatively compared with the partial shunt group. Conclusions: The partial portacaval shunt effectively controls variceal hemorrhage. Compared with direct side-to-side portacaval shunt, partial shunt preserves long-term hepatic function and minimizes postoperative encephalopathy. We conclude that the partial portacaval shunt is the preferred approach over direct shunts for patients with cirrhosis and with variceal bleeding. (Surgery 2000;127:614-21.)