Our laboratory has developed virtually identical techniques for the isolation and culture of mammary epithelial cells (MEC) from rats and humans. In a cell-mediated mutagenesis assay, rat MEC activated 7,12-dimethylbenz(a)anthracene (DMBA) but not benzo(a)pyrene [B(a)P] to mutagenic forms, and the opposite pattern was found with human MEC. These species-specific patterns were not readily explained by either qualitative or quantitative differences in Phase I metabolism of these compounds. In contrast, relative levels of covalent binding of these compounds to DNA in the human and rat cells under identical assay conditions generally parallel the pattern of the mutagenesis results, while not reflecting the absolute levels of metabolism in each system. The ability of the rat MEC to bind relatively higher levels of DMBA than B(a)P to nuclear DNA, and the reversed pattern in human MEC, was found at all incubation times tested between 6 and 48 h. Culture density was found to exert a greater effect on the levels of PAH-DNA binding in rat than in human cells, but in neither case did it affect the ratio of DMBA to B(a)P binding within a species. C2SO4 gradient separation of nuclear macromolecules from PAH-treated MEC revealed that the relative DNA binding levels of DMBA and B(a)P did not correlate with relative levels of nuclear protein binding. For both species, nuclear (DNA + protein) binding levels of B(a)P were approximately 2-fold higher than DMBA. However, these binding levels were 4 to 5-fold higher for both carcinogens in the human than in the rat MEC. The species-specific patterns of PAH activation shown by these cells suggest that caution should be used in extrapolating rodent carcinogenesis data to humans, for either quantitative or qualitative purposes.