A commercial polybrominated biphenyl mixture was separated chromatographically into two fractions on neutral alumina (A A and A B) and into three fractions on Florisil (F A, F B, and F C) by sequential elution with solvents of increasing polarity. Using established methods, the activity of each fraction as hepatic microsomal cytochrome P-450- and/or cytochrome P-448-dependent monooxygenase enzyme inducers was examined in the male Wister rat. Like the coadministration of phenobarbitone and 3-methylcholanthrene, commercial PBBs, either unfractionated or reconstituted from its various fractions, induced both cytochromes P-450 and P-448. Both cytochromes were also induced by the less-polar fractions A A and F A. In contrast, little or no inductive effects were exhibited by the more polar Florisil fractions, F B and F C, indicating that the ability of commercial PBBs to induce cytochrome P-448 is not due to contaminating brominated dibenzofurans or dibenzodioxins. Unlike the polar Florisil fraction, the more polar alumina fraction, A B, was a potent microsomal enzyme inducer. This fraction was enriched in 2,3,3′,4,4′,5-hexa- and 2,2′,3,3′,4,4′,5-heptabromobiphenyl and also contained unassigned monochloro derivatives of a penta- and hexabromobiphenyl, namely C 12H 4Br 5Cl and C 12 H 3Br 6Cl, respectively. The data strongly suggest that the biologic effects of the commercial polybrominated biphenyl mixture are due to the various halogenated biphenyls present. These results are discussed in terms of the reported toxic potency of each PBB fraction and with reference to the known biologic activity of individual polybrominated biphenyl congeners or their chloro analogs.