Two sets of prognostic indicators were recently proposed for selection of patients with acute liver failure for emergency liver transplantation. According to the London criteria, patients with non-paracetamol-induced acute liver failure should be referred for liver transplantation when the prothrombin time is > 100 s or when any three of the following prognostic indicators are present: age < 10 or > 40 yr; non-A, non-B hepatitis, halothane hepatitis or idiosyncratic drug reaction; duration of jaundice before onset of encephalopathy > 7 days; prothrombin time > 50 s; serum bilirubin > 300 mumol/l. According to the Clichy criteria, in acute viral hepatitis, liver transplantation should be decided in patients with coma or confusion, and Factor V < 20% (age < 30 yr) or < 30% (age > 30 yr). To assess the accuracy of these criteria, 81 non-transplanted patients with non-paracetamol-induced acute liver failure were retrospectively studied. The mortality rate was 0.81. The predictive accuracies, respectively on admission and 48 h before death, were 0.80 and 0.79 for the London criteria, and 0.60 and 0.73 for the Clichy criteria. The positive and negative predictive values, 48 h before death, were 0.89 and 0.47 for the London criteria, and 0.89 and 0.36 for the Clichy criteria, respectively. In the 49 patients with acute viral liver failure, the results of the Clichy criteria were similar. In a subgroup of 24 patients who had not received either fresh frozen plasma or sedative-hypnotic drug, the positive predictive values were equal to 1 for the two sets of prognostic indicators, but the predictive accuracies only slightly increased.(ABSTRACT TRUNCATED AT 250 WORDS)