Phenanthrene, generally considered to be a non-carcinogen, was converted by mammalian tissue preparations to products that were mutagenic for Salmonella typhimurium TA100 and TA1537. In TA100 the mutagenic response was highly dependent on the activation system used. High amounts of 9000 x g supernatant fraction from the liver of rats induced by Aroclor 1254 were required. Equivalent amounts of microsomal or cytosolic fraction alone did not activate phenanthrene to an observable extent. Furthermore, this activation was only observed when the rats had been treated with Aroclor. Liver preparations from control rats and from rats treated with phenobarbital, beta-naphthoflavone, a mixture of both, and transstilbene oxide failed to activate phenanthrene to mutagens for TA100. Interestingly, liver microsomes and 9000 x g supernatant fractions of Aroclor-treated mice also failed significantly to activate phenanthrene to mutagens for this strain. Addition of pure epoxide hydrolase to the S9 mix had no influence on this activation. Glutathione (GSH) decreased the mutagenicity, but uridine diphosphate glucuronic acid (UDPGA) had only minor effects. An adenosine-3'-phosphate-5'-sulfate phosphate (PAPS) generating system, however, increased the number of his+ revertants from TA100 (2.7-fold). TA1537 was reverted by mutagens produced from phenanthrene by liver microsomes or 9000 x g supernatant fraction, when the microsomal epoxide hydrolase was inhibited by 1,1,1-trichloropropene oxide. This activation pathway exists in Aroclor-treated rats and mice. The results show that at least 2 different pathways for metabolic activation of phenanthrene exist which were observed in 2 differentially sensitive tester strains and distinguished by their different metabolic requirements. Furthermore, the study shows that earlier suggestions do not hold that equivalent results can be obtained by inducing animals with a combination of phenobarbital and beta-naphthoflavone instead of the environmentally persistent Aroclor 1254. Moreover, the study provides a striking example that the use of 9000 x g supernatant in amounts corresponding to standard practice but sub-optimal for a particular compound only impede the detection of a weak mutagen and that the rapid inactivation of active metabolites by inactivating enzymes may be responsible for negative results in mutagenicity testing.