Peritoneovenous shunts (PVSs) have provided salutary effects on medically recalcitrant ascites, functional renal impairment, nutritional derangements, ventilatory embarrassment, and locomotion potential in patients with cirrhosis. While the LeVeen (LPVS) and Denver (DPVS) PVSs are most frequently implanted in such patients, postoperative complications of bleeding gastroesophageal varices, sepsis, and shunt occlusion occur with notable frequency. Addressing primarily the complication of PVS occlusion, a randomized prospective trial of LPVSs and DPVSs was conducted in cirrhotic patients with refractory ascites. From July 1, 1982 to July 1, 1984, 26 initial PVSs were implanted for hepatic-related intractable ascites. Twenty-two patients were eligible for randomization (cirrhosis, sterile ascites, initial PVS, total bilirubin level ≦6.0 mg/dL, prothrombin time ≦5-s prolongation, serum creatinine level ≦2.0 mg/dL [creatinine clearance rate ≧20 mL/min], absence of recent [<30 days] bleeding gastroesophageal varices, or absent spontaneous encephalopathy). Twelve LPVSs and ten DPVSs were implanted; however, one patient with a DPVS was found to have hepatic polycystic disease and was excluded from analysis. All patients were followed up until death or January 1, 1985. The PVS patency determinations included contrast shuntography, technetium Tc 99m albumin scintigraphy, sequential manual compression (DPVS), and operative or autopsy observation. Using the Kaplan-Meier actuarial analysis, the LPVS patency proved to be highly superior to that of the DPVS, while survival was not significantly different. As LPVS and DPVS complications other than patency are comparable, the LPVS is preferred for its superior patency in cirrhotic patients with intractable ascites.