A multicenter longitudinal study was performed to assess the survival of hepatitis B surface antigen positive compensated cirrhosis, primarily in relation to hepatitis B virus replication and hepatitis delta virus infection, and to construct a prognostic index based on entry characteristics. This cohort study involved nine university medical centers in Europe. Three hundred and sixty-six Caucasian HBsAg positive patients with cirrhosis who had never had clinical manifestations of hepatic decompensation were enrolled and followed for a mean period of 72 months (6 to 202 months). Inclusion criteria were biopsy-proven cirrhosis, information on serum hepatitis B e antigen and antibody to hepatitis D virus at the time of diagnosis and absence of complications of cirrhosis. At entry 35% of the patients were HBeAg positive, 48% of the patients tested were HBV-DNA positive and 20% anti-HDV positive. Death occurred in 84 (23%) patients, mainly due to liver failure (45 cases) or hepatocellular carcinoma (23 cases). The cumulative probability of survival was 84% and 68% at 5 and 10 years, respectively. Cox's regression analysis identified six variables that independently correlated with survival: age, albumin, platelets, splenomegaly, bilirubin and HBeAg positivity at time of diagnosis. According to the contribution of each of these factors to the final model, a prognostic index was constructed that allows calculation of the estimated survival probability. No difference in survival of hepatitis D virus infected and uninfected patients was observed. Termination of hepatitis B virus replication and/or biochemical remission during follow up correlated with a highly significant better survival. These data show that in compensated cirrhosis B, hepatitis B virus replication, age and indirect indicators of poor hepatic reserve and established portal hypertension significantly worsen the clinical course of the disease, whereas hepatitis D virus infection does not influence the prognosis. The highly significant improvement in life expectancy following cessation of hepatitis B virus replication and biochemical remission favors antiviral therapy in those patients with a guarded prognosis, as estimated by a prognostic index.