In 1989, we reported on the efficacy of liver transplantation in primary biliary cirrhosis (PBC) by demonstrating that the actual patient survival following transplantation was significantly better than without transplantation as predicted by a mathematical survival model ("Mayo natural history model"). Our aim in this investigation was to determine an optimal time to perform liver transplantation in PBC. One hundred forty-three patients with PBC undergoing liver transplantation were followed prospectively. Disease severity was measured immediately before transplantation by a summary score ("risk score") used in the Mayo natural history model, namely age, bilirubin, albumin, prothrombin time, and the presence or absence of edema. Proportional hazards analyses were performed assessing patient survival following transplantation. The influence of disease severity immediately pretransplantation on resource utilization for liver transplantation was assessed. Compared with our report in 1989, liver transplantation was performed at an earlier stage of disease (e.g., median risk score: 7.5 vs. 8.3; P < .01). Following transplantation, patient survival probabilities at 1, 2, and 5 years were 93%, 90%, and 88%, respectively. In the proportional hazards analysis, the risk of death following transplantation remained low until reaching a risk score of 7.8. In contrast, risk scores greater than 7.8 were associated with a progressively increased mortality. Resource utilization measured by the days in the intensive care unit (ICU) and hospital and the requirement for intraoperative blood transfusions was significantly greater in recipients who had higher risk scores before transplantation. Our data suggest that an optimal timing for liver transplantation, as determined by patient survival and resource utilization, appears to be at a risk score around 7.8 in patients with PBC.