In a continuation of a trial for which preliminary results were reported in the Journal two years ago, a total of 64 patients with Child Class C cirrhosis and variceal hemorrhage requiring six or more units of blood were randomly assigned to receive either a portacaval shunt (32 patients) or endoscopic sclerotherapy (32 patients). The duration of initial hospitalization and the total amount of blood transfused during hospitalization were significantly less in the patients receiving sclerotherapy (P less than 0.001). There was no difference in short-term survival (50 percent of the sclerotherapy group were discharged alive, as compared with 44 percent of the shunt-surgery group). Both groups were followed for a mean of 530 days after randomization. Rebleeding from varices, the duration of rehospitalization for hemorrhage, and transfusions received after discharge were all significantly greater in the sclerotherapy group (P less than 0.001). Forty percent of the sclerotherapy-treated patients discharged alive (7 of 16 patients) ultimately required surgical treatment for bleeding varices, despite a mean of 6.1 treatment sessions. Health care costs and long-term survival did not differ significantly between the groups (P greater than 0.05). We conclude that although endoscopic sclerotherapy is as good as surgical shunting for the acute management of variceal hemorrhage in poor-risk patients with massive bleeding, sclerotherapy-treated patients in whom varices are not obliterated and bleeding continues should be considered for elective shunt surgery.