Ashby et al. (Carcinogenesis 1:1-7, 1980) have reported that 4-N-pyrrolidinylazobenzene is a powerful toxin and transforming agent in BHK cells. They have also proposed a scheme for the prediction of carcinogenic activity which is based on structural analysis of this and other compounds, including N-methyl-5-phenylazoindoline, tested in their BHK system. Contrary to their prediction, we found that 4-N-pyrrolidinylazobenzene and N-methyl-5-phenylazoindoline were not carcinogenic when fed to adult rats under conditions where N,N-dimethyl-4-aminoazobenzene induced a high incidence of hepatic carcinomas. 4'-Ethyl substitution of the pyrrolidine derivative yielded a weak carcinogen. These data suggest that complex structure-activity analyses for carcinogenic activity in this class of compounds are premature in the absence of data on whole animals. The transformation of BHK cells appears to require further validation as a predictor of the carcinogenic activity of chemicals for animals.