Aim: To retrospectively analyze different methods in differentiating biliary atresia from idiopathic neonatal hepatitis. Methods: Sixty-nine infants with cholestatic jaundice and final diagnosis of idiopathic neonatal hepatitis (INH) and biliary atresia (BA) were studied retroprospectively from January 2004 to December 2006. A thorough history and physical examination were undertaken. All cases underwent abdominal magnetic resonance cholangiography (MRCP), ultrasonography (US), hepatobiliary scintigraphy (HBS), HBS single-photon emission computer tomography (HBS SPECT), and operation or percutaneous liver biopsy. The accuracy, sensitivity, specificity, and predictive values of these various methods were compared. Results: There were 39 girls and 30 boys, among whom 35 had INH (age, 61 +/- 17 days) and 34 had BA (age, 64 +/- 18 days). The mean age at onset of jaundice was significantly lower in cases of BA when compared to INH cases (9 +/- 13 vs. 20 +/- 21 days; P=.032). The diagnostic accuracy of different methods was as follows: liver biopsy, 97.1%; HBS SPECT, 91.30%; MRCP, 71.01%; HBS, 66.67%; US, 65.22%. Conclusion: Our results indicate that biopsy of the liver is considered as the most reliable method to differentiate INH from BA. The accuracy of HBS SPECT is higher than that of MRCP, HBS, and US. There was no significant difference in diagnostic accuracy among MRCP, HBS, and US. (C) 2009 Published by Elsevier Inc.