Cyclopenta[cd]pyrene (1) and its congeners dicyclopenta [cd,mn]-(2), dicyclopenta [cd,fg]-(3), dicyclopenta [cd,jk]pyrene (4), which were all identified as constituents of combustion exhausts, as well as their partially hydrogenated derivatives 3,4-dihydrocyclopenta[cd]-(5), 1,2,4,5-tetrahydrodicyclopenta[cd,mn]-(6), 5,6,7,8-tetrahydrodicyclopenta[cd,fg]- (7) and 1,2,6,7-tetrahydrodicyclopenta[cd,jk]pyrene (8), were assayed for mutagenicity in the Salmonella typhimurium strain TA98 using different concentrations of microsomal protein in the metabolic activation system (S9-mix, with S9-fraction from liver of Aroclor-1254-treated rats: 2, 4 and 10% (v/v), respectively). Whereas a positive mutagenic response is found for 1-4 in the presence of S9-mix, 5-8 exert no mutagenicity either with or without S9-mix. Since for 1-4 the highest response is observed with S9-mix 2% (v/v) instead of the standard 4% (v/v), a one-step activation pathway, i.e. epoxidation of the five-membered ring olefinic bonds, appears to be operational. Surprisingly, 3 and, to a lesser extent, 2 (11.7 versus 4.2 His revertants/nmol) also give a positive response in the absence of S9-mix. Hence, 2 and 3 are expected to contribute to the direct-acting mutagenicity of the non-polar fraction of combustion exhausts. Presumably for the direct-acting mutagenicity one-electron transfer processes play a role in bioactivation. The experimental observations are supported by semi-empirical AM1 calculations on the possible ultimate metabolites, i.e. mono-epoxides (2a-4a), cis-di-epoxides (2b-4b) and trans-di-epoxides (2c-4c) and the related mono-hydroxy carbocations (2d-4d and 2e-4e), and the radical anions 1 •− -4 •− .