To assess the reliability of technetium-99m disofenin scanning in evaluating neonatal cholestasis, 33 neonates (less than 3 months of age) with direct hyperbilirubinemia were evaluated prospectively by cholescintigraphy. Results of this test were compared to those of standard serum tests of liver function, ultrasonography, and liver biopsy. The diagnosis of biliary atresia was suggested by a serum gamma-glutamyl transpeptidase (gamma-GTP) greater than 300 units/L, absence of the gallbladder on ultrasonography, and a lack of detectable radioisotope in the gastrointestinal and/or extrahepatic biliary tract on cholescintigraphy. Each of these tests lacked sensitivity and/or specificity when compared to liver biopsy. Of the nine neonates with biliary atresia, three had gallbladders identified by ultrasonography and two had gamma-GTP less than 300 units/L. Of the 24 neonates without biliary atresia, eight had cholescintigraphy without detectable radioisotope excretion, four had ultrasonography that failed to visualize the gallbladder, and nine had gamma-GTP greater than 300 units/L. Cholescintigraphy excluded the diagnosis of biliary atresia when gut and/or extrahepatic biliary excretion of isotope was seen. However, cholescintigraphy required more time, 6-8 days, and was less specific than ultrasonography and liver biopsy. We recommend that cholescintigraphy should not be routinely used in evaluating neonatal cholestasis, especially if it delays surgical intervention.