Coumarin produces histological liver damage in the rat and dog, but not, following an intake of up to 67·5 mg/kg/day for 2 years, in baboons. Grave doubts have been expressed over the acceptability of the diagnosis of 'bile-duct carcinomas' in rats fed coumarin at 5000 or 6000 ppm in the diet for 2 years, especially in view of the absence of significant metastasis in these animals and of the negative carcinogenic findings obtained in an earlier study in rats. Coumarin does not act as a co-carcinogen in mouse skin and no evidence of carcinogenicity was obtained following repeated subcutaneous administration to rats in a limited experiment. Since the rat metabolizes coumarin quite differently from man, the suitability of the rat as a test species in predicting the hepatotoxic risk to man is questionable. Whether coumarin or a metabolite is the active hepatotoxin has not been established conclusively, but the evidence suggests that metabolism is an important factor in determining the hepatotoxic response. Therefore species differences in metabolism should be taken into account in evaluating the hepatotoxic hazard to man. © 1979.