Emergency portacaval shunt for variceal bleeding is associated with a high operative mortality, particularly if used as a last resort. Because of this, a strong case has been made against emergency shunt. This report describes an experience with emergency portacaval shunt for the treatment of variceal bleeding when used systematically after hemodynamic stabilization and control of the bleeding episode with balloon tamponade, if necessary, in patients with mild or moderate liver disease. The population studied comprised 62 consecutive patients who rebled from varices while participating in a controlled trial of propranolol for the prevention of rebleeding. Of the 62 patients, nine died of massive hemorrhage and 53 survived the hemorrhage. Of the 53 survivors, 11 had severe liver disease and were not considered for shunt surgery. Of the remaining 42 patients with mild or moderate liver disease, 36 had emergency central portacaval shunt. The interval between endoscopic diagnosis of variceal bleeding and surgery averaged 19 +/- 3 hours (mean +/- SE). The operative mortality rate, defined as in-hospital mortality, was 19%. One- and 2-year survival rates were 78% and 71%, respectively. The incidence of postoperative hepatic encephalopathy was 36%; all patients responded favorably to protein restriction and lactulose. Thus, under specific conditions, emergency portacaval shunt results in an acceptable long-term survival rate. In patients with mild or moderate liver disease, emergency portacaval shunt should be considered when other forms of treatment for the prevention of variceal rebleeding have failed.