This study was conducted to determine the rate of detection of serum hepatitis B virus (HBV) DNA in hepatitis B surface antigen (HBsAg)-negative decompensated cirrhotic patients who had hepatitis B core a n d or surface antibodies (anti-HBc andor anti-HBs), and to compare the outcome of HBsAg-positive cirrhotic patients who did or did not clear HBsAg during followup. Six (5%) of 121 HBsAg-positive cirrhotic patients lost HBsAg after 0.2 to 17.1 years (mean, 9.1 -+ 6.2 yr) of followup. The cumulative rates of loss of HBsAg at 1,5,10, and 15 years were, respectively, 1.3%, 1.3%, 7.4%, and 44.5%. Compared with the patients who remained HBsAg-positive, those who lost HBsAg had milder disease at presentation and significantly longer survival. Of the patients who lost HBsAg, 83% had improvement in liver function after the loss of HBsAg, and all were alive at the time of writing (0.8 to 5.7 years after loss of HBsAg), whereas 27% of those who remained HBsAg-positive had died and 2Wo had deterioration in liver function. The rate of detection of serum HBVDNA by polymerase chain reaction (PCR) assay was higher in HBsAg-positive cirrhotic patients who lost HBsAg: 67% versus cirrhotic patients who had no previous history of chronic HBV infection; 16% (cryptogenic) and 29% (hepatitis C virus andor alcoholinduced liver disease). In summary, we found that using PCR, serum HBV DNA can be detected in 28% of HBsAgnegative cirrhotic patients who were studied, but the pathogenic significance of such small amounts of virus is not clear. Liver function can improve and survival can be prolonged in HBsAg positive decompensated cirrhotic patients who subsequently lost HBsAg. (HEPATOL-