Despite the efficacy of shunt surgery in the treatment of variceal bleeding, less effective nonoperative therapies are being substituted because surgical shunt does not modify survival and increases hepatic encephalopathy. However, the real impact of shunt surgery on the natural history of ascites and its complications has not been established. The course of 204 Child-Pugh A and B cirrhotic patients with variceal bleeding included in three controlled trials of our units who survived first hospitalization was updated. Ninetyeight patients (group I) were treated by portacaval (56 patients) or distal splenorenal (42 patients) shunt, whereas 106 (group II) were treated by nonshunt procedures (sclerotherapy in 89 patients and staple transection in 17 patients). As expected, the 5-yr probability of variceal rebleeding was lower (13% vs. 44%) and hepatic encephalopathy higher (50% vs. 28%) in group I than in group II, and survival was similar (67% vs. 60%). Shunt surgery had a great impact on the natural history of ascites and its complications. The probability of occurrence of ascites (15% vs. 73%; p < 0.0001), spontaneous bacterial peritonitis (2% vs. 21%; p < 0.0001) and hepatorenal syndrome (4% vs. 21%; p < 0.01) was greatly reduced. These beneficial effects accounted for the lower percentage patients requiring readmissions (51% vs. 70%; p = 0.02) and shorter total time spent in hospital (14 + 22 vs. 26 + 39 days/patient p = 0.01) in group I. These data indicate that shunt surgery, in addition to reducing the probability of rebleeding, markedly decreases the probability of ascites, spontaneous bacterial peritonitis and hepatorenal syndrome development. Therefore, ascites and its complications should be considered in any therapeutic trial assessing the efficacy of surgical or percutaneous portacaval anastomosis on variceal bleeding. (HEPATOLOGY 1994;20:584-591).