To establish principles for choosing between endoscopic retrograde and percutaneous transhepatic cholangiography, we randomized 52 consecutive jaundiced patients with clinically suspected obstructive jaundice. The bile ducts were visualized in 85% by the endoscopic and in 84% by the transhepatic route. A conclusive diagnosis was reached in 89% and 68% of the patients, respectively, but the difference is not significant (0.10 less than P less than 0.20). If the planned type of cholangiography failed, the other method was tried. By comparing the total numbers of investigations, a conclusive diagnosis was achieved in 91% by endoscopic and 69% by transhepatic route, which is significantly different (P less than 0.05). We therefore prefer to do endoscopic cholangiography initially with transhepatic cholangiography as the complementary method.