Apolipoprotein A-I (apo A-I) has been found to be decreased in adults with cirrhosis, but it has not been routinely used for prognostic purposes thus far. This study was performed to determine apo A-I levels in childhood cirrhosis and to establish some prognostic cut-off values. Apo A-I levels of 78 children with chronic liver disease, 38 of whom had cirrhosis as well, were studied. Mean values of cirrhotic, non-cirrhotic and healthy children were not different (p > .05). However, in cirrhotic children, the highest value was detected in the Child-Pugh A group, and it was different from those of the B and C groups (p < .05 and p < .001, respectively). Apo A-I was the lowest in the moderate risk group of Malatack's model, and was significantly different from the low risk group (p < .05). Apo A-I was inversely correlated with Malatack score, Child-Pugh score, total bilirubin, conjugated bilirubin, and prothrombin time (p < .01, p < .01, p < .01, p < .01, p < .05, respectively). In cirrhotic children with cholestasis, apo A-I was lower than in non-cholestatic children (p < .05). Apo A-I value < 80 mg/dl had 84% specificity and 84% negative predictive value for the high risk group of Malatack's model. Similarly, Apo A-I value < 83 mg/dl had 95% specificity and 87% negative predictive value for the Child-Pugh C group. We concluded that Apo A-I is a sensitive and specific parameter of poor prognosis in childhood cirrhosis.