The hepatotoxic effects of cholelitholytic bile acids, ursodeoxycholic and chenodeoxycholic acids, were compared with each other and with those of lithocholic acid, a known hepatotoxic bile acid, in the rabbit. Male New Zealand white rabbits were fed regular laboratory chow containing ursodeoxycholic, chenodeoxycholic, or lithocholic acid at a concentration of 0.5 per cent (w/w) for 14 days. The control group was fed the chow without added bile acids. The mortality rate was highest (six of 12) in the lithocholate group, intermediate (two of eight) in the chenodeoxycholate group, and lowest (none of six) in the ursodeoxycholate group. Light microscopy of the liver revealed fibrosis, inflammation, and bile duct proliferation in the portal regions in the three experimental groups; however, the lesions in the lithocholate and chenodeoxycholate groups were more severe and often associated with periportal extension of fibrosis and focal necrosis of the parenchyma. In addition, electron microscopy revealed distortion of bile canaliculi, conspicuous bundles of intermediate-sized filaments, expansion of pericanalicular cytoplasmic matrix due to apparent accumulation of microfilaments, prominence of lysosomes, and fragmentation of cisternae of the rough endoplasmic reticulum. These ultrastructural changes were less marked and often absent in the ursodeoxycholate group. The serum L-alanine aminotransferase activity increased 5- to 6-fold in the lithocholate and chenodeoxycholate groups, whereas it remained less than 2-fold of the control level in the ursodeoxycholate group on day 14. The serum lithocholate concentration was markedly elevated to comparable levels in all three groups, whereas ursodeoxycholate was highly increased in the ursodeoxycholate group but undetectable in the other groups at the time of sacrifice. It is concluded that (1) although the oral administration of three bile acids induces hepatic injuries in the rabbit, ursodeoxycholate causes less severe injury than do the other two, (2) the advantage of ursodeoxycholate versus chenodeoxycholate is probably relative rather than absolute, (3) lithocholate formed through metabolic conversion from ursodeoxycholate may be responsible for the most part for hepatotoxicity, and (4) it is possible that the concurrent presence of ursodeoxycholate may mitigate lithocholate's hepatotoxicity.