Tumorigenic activities of the (+)- and (-)-enantiomers of the diastereomeric, bay-region benzo(c)phenanthrene 3,4-diol-1,2-epoxides were evaluated in two mouse tumor models. In an initiation-promotion experiment on mouse skin, a single topical application of 10, 25, or 75 nmol of the compounds was followed by 20 weeks of promotion with 12-O-tetradecanoylphorbol-13-acetate. Of the four optical isomers of the bay-region diol epoxides, (-)-(R,2S,3S,4R)-3,4-dihydroxy-1,2-epoxy-1,2,3,4-tetrahydrogenzo(c )phenanthrene [(-)-diol epoxide-2] and (+)-(1R,2S,3R,4S)-3,4-dihydroxy-1,2-epoxy-1,2,3,4-tetrahydrobenzo(c) -phenanthrene [(+)-diol epoxide-1] had equally high tumor-initiating activity while (+)-[1S,2R,3R,4S]-3,4-dihydroxy-1,2-epoxy-1,2,3,4-tetrahydrobenzo (c)phenanthrene [(+)-diol epoxide-2] had less than one-half of the activity of (-)-diol epoxide-2 and (+)-diol epoxide-1. (-)-(1S,2R,3S,4R)-3,4-Dihydroxy-1,2-epoxy-1,2,3,4-tetrahydrobenzo(c) -phenanthrene [(-)-diol epoxide-1] was inactive at the doses tested. In newborn mice, (-)-diol epoxide-2 was almost 10-fold more active in producing lung tumors (average number of lung tumors/mouse) than the next most active compound, (+)-diol epoxide-2, at a total dose of 10 nmol. The enantiomers of diol epoxide-1 were inactive at this dose. When the total dose of each optical isomer was increased to 50 nmol, (-)-diol epoxide-1 was still inactive, and (+)-diol epoxide-1 produced a significant number of lung tumors (0.9 lung tumor/mouse), but this isomer still had less than 10% of the activity of the (+)- and (-)-diol epoxide-2 isomers. (-)-Diol epoxide-2, but none of the other optical isomers, also produced a significant incidence of hepatic tumors at the higher dose, and this compound was found to be the most tumorigenic bay-region diol epoxide ever tested in newborn mice. Racemic diol epoxide-1 had approximately 1% of the tumorigenic activity of racemic diol epoxide-2 in newborn mice, but both racemates had equal tumor-initiating activity on mouse skin. These results dramatically illustrate the complexities involved in ranking the relative tumorigenic activities of compounds in different tumor models.