BACKGROUND : Biliary atresia is a common cause of cholestasis. In our experience, patients with biliary atresia are referred late, with the diagnosis based on an absent gall bladder at ultrasonography. Such late referrals may render patients inoperable and not acceptable for formal intraoperative diagnosis. OBJECTIVES : To determine the usefulness of an absent gall bladder on ultrasonography, and of biochemical features, in differentiating biliary atresia from other causes of cholestasis, using liver needle biopsy as a gold standard. METHODS : A retrospective file review of 150 infants presenting with cholestasis to Steve Biko Academic Hospital Paediatric Gastroenterology and Hepatology Unit from January 2008 to August 2014 was undertaken. Clinical, serum biochemical, abdominal ultrasonography and liver histology findings were analysed. Three groups were compared, based on liver histology findings, consisting of patients with biliary atresia, neonatal hepatitis, and other diagnoses, respectively. RESULTS : A total of 66/150 patients had biliary atresia, based on liver histology findings of extrahepatic obstruction. Their mean age was 4.7 (2.9) months, higher than in the other groups. In those with biliary atresia, the age at diagnosis, splenomegaly, and gamma glutamyl transferase (GGT), aspartate transaminase (AST) and GGT/AST ratio values were significantly different from the other groups. Total and conjugated bilirubin levels were similar among the groups. Ultrasonography was 69.7% sensitive, 98.8% specific and had a positive predictive value of 97.9% for biliary atresia. Ultrasonography missed 30.3% of patients with biliary atresia. CONCLUSION : Ultrasonography has poor sensitivity but good specificity in screening for biliary atresia; however, other investigations are necessary to confirm the diagnosis.