Variceal hemorrhage continues to be a major cause of morbidity and mortality in cirrhotic patients. Transjugular intrahepatic portosystemic shunt (TIPS) is gaining wide acceptance as a treatment for several complications of portal hypertension. The aim of the current randomized study was to compare the transjugular shunt and endoscopic sclerotherapy (ES) for the prevention of variceal rebleeding (VB) in cirrhotic patients. Forty-six consecutive cirrhotic patients with variceal bleeding were randomly allocated to receive either transjugular shunt (22 patients) or ES (24 patients) 24 hours after control of bleeding. VB (50% vs. 9%) and early (first 6 weeks) VB (33% vs. 5%) were significantly more frequent in sclerotherapy patients; the actuarial probability of being free of VB was higher in the shunt group (P <.002). Eight patients (33%) of the sclerotherapy group and 3 patients (15%) of the shunt group died; the actuarial probability of survival was higher for the shunted patients (P <.05); 6 patients in the sclerotherapy group and none in the shunt group died from VB (P <.05). No difference was found in the proportion of patients with clinically evident hepatic encephalopathy (HE). These results show that the transjugular shunt is more effective than sclerotherapy in the prevention of both early and long-term VB. Moreover, a significant improvement in survival was found in the shunt group.