Biliary atresia (BA) is very difficult to distinguish from neonatal hepatitis (NH) and its prognosis depends on the age at the time of Kasai operation. Therefore early differentiation between these two conditions is very important. Although various clinical and laboratory tests have been reported to differentiate between them, they are still of limited value. From 1980 to 1999, forty-five infants were referred to our pediatric surgical unit for operation for suspected BA. Eight patients underwent Kasai operation immediately because late diagnosis. These were excluded from the study. The clinical history, physical findings, radiologic and laboratory examinations of 37 cases were analyzed retrospectively. The average age of BA (n=20) was 55.1+/-6.7 days, and that of NH (n=17) was 55.8+/-5.6 days. The sex ratio of BA was 13:7, and that of NH was 14:3. All the patients had obstructive jaundice and acholic stool except 4 BA and 6 NH patients. Acholic stool with yellow component was more frequent in NH. Onset of jaundice was within 2 weeks after birth in 85% of BA, and in 65% of NH. The onset of acholic stool was within 2 weeks after birth in 60% of BA, and in 23.5% of NH. The duration of jaundice and acholic stool of BA were 50.9+/-6.6 days and 41.3+/-8.4 days and those of NH were 40.1+/-3.1 days and 26.6+/-5.4 days respectively. The ultrasonogram and hepatobiliary scan were useful, but not a definitively reliable method for the differentiation of these two diseases. There was no difference in laboratory data. Seventeen cases had NH among 45 referred cases for Kasai operation with the clinical impression of BA, and 4 cases of 17 NH cases needed to be explored to rule out BA. In conclusion, false positive rate of clinical impression of BA was 37.8%, and negative exploration rate was 8.9%. Therefore, careful clinical observation for 1-2 weeks by an experienced pediatric surgeon was very important to avoid unnecessary operation to rule out NH up to the age of 8 - 10 weeks, so long as the stool had yellow component.