We report herein the results of extended follow-up of an expanded randomized clinical trial comparing transjugular intrahepatic portosystemic shunt (TIPS) to 8 mm prosthetic H-graft portacaval shunt as definitive treatment for variceal bleeding due to portal hypertension. Beginning in 1993, through this trial, both shunts were undertaken as definitive therapy, never as a "bridge to transplantation." All patients had bleeding esophageal/gastric varices and failed or could not undergo sclerotherapy/banding. Patients were excluded from randomization if the portal vein was occluded or if survival was hopeless. Failure of shunting was defined as inability to shunt, irreversible shunt occlusion, major variceal rehemorrhage, hepatic transplantation, or death. Median follow-up after each shunt was 4 years; minimum follow-up was 1 year. Patients undergoing placement of either shunt were very similar in terms of age, sex, cause of cirrhosis, Child's class, and circumstances of shunting. Both shunts provided partial portal decompression, although the portal vein-inferior vena cava pressure gradient was lower after H-graft portacaval shunt (P < 0.01). TIPS could not be placed in two patients. Shunt stenosis/occlusion was more frequent after TIPS. After TIPS, 42 patients failed (64%), whereas after H-graft portacaval shunt 23 failed (35%) (P < 0.01). Major variceal rehemorrhage, hepatic transplantation, and late death were significantly more frequent after TIPS (P < 0.01). Both TIPS and H-graft portacaval shunt achieve partial portal decompression. TIPS requires more interventions and leads to more major rehemorrhage, irreversible occlusion, transplantation, and death. Despite vigilance in monitoring shunt patency, TIPS provides less optimal outcomes than H-graft portacaval shunt for patients with portal hypertension and variceal bleeding.