The effect of azathioprine on survival of patients with primary biliary cirrhosis was studied prospectively in a multinational, double-blind, randomized clinical trial including 248 patients of whom 127 received azathioprine and 121 placebo. There were 57 deaths in the azathioprine group and 62 in the placebo group. The actual survival was slightly longer during azathioprine than during placebo treatment. Using Cox multiple regression analysis and adjusting for slight imbalance between the two treatment groups, the therapeutic effect of azathioprine was statistically significant (p = 0.01), with azathioprine reducing the risk of dying to 59% of that observed during placebo treatment (95% confidence interval 40%-90%) or improving survival time by 20 mo in the average patient. Furthermore, azathioprine slowed down progressing incapacitation. Side effects of azathioprine were relatively few. The analysis revealed that the following five variables independently implied poor prognosis: high serum bilirubin, old age, cirrhosis, low serum albumin, and central cholestasis. These factors were combined to a "prognostic index" for prediction of outcome in new patients. The index was validated on independent patient data. On the basis of these results we recommend azathioprine as a routine treatment of primary biliary cirrhosis.