Using an improved set of lactose-auxotrophic Escherichia coli tester strains, the proportion of the six possible transitions and transversions after mutagen exposure was assessed. Mutagenic specificity was determined in plate-incorporation assays using lactose-containing minimal medium for the selection of revertants, either after application of directly acting mutagens or by including a metabolic activation system with rat liver S9-extract. The differential and dose-dependent response of the six tester strains was shown by treating the bacteria with described diagnostic mutagens and other directly DNA damaging substances, e.g., N-methyl-N-nitrosoguanidine (MNNG) and benzo[a]pyrene-diolepoxide (BPDE). Polycyclic aromatic hydrocarbons and aromatic amines were investigated in the presence of an external metabolic activation system. Benzo[a]pyrene (BaP) yielded similar mutation profiles as its ultimate mutagen BPDE, if 100-fold increased doses were applied. In contrast to the mutation profile of BaP, which was dominated by G:C-T:A transversions, mutagenesis with benzo[c]phenanthrene (BcPh) produced predominantly A:T-T:A transversions. The same base change was observed with 5-methylchrysene and found to be missing with 5,6-dimethylchrysene, while both compounds caused G:C-A:T transitions. The aromatic amines 4-aminobiphenyl (4-ABP), 2-aminoanthracene (2-AA) and 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhlP) yielded similar yet distinguishable mutation profiles. Base-substitution reversion profiles of the chemical mutagens were in agreement with those obtained in other systems and with molecular analysis of mutants induced by these agents.