With the aim of optimizing the efficiency of S9 fractions used in in vitro mutagenicity assays, different schemes for the induction of liver enzymes in rats were tried and the amount of S9 fraction required was assessed. The activity of 2-anthramine (2AA), 2-acetylaminofluorene (2AAF), 3-methylcholanthrene (3MTCL) and benzo[a]pyrene in bacterial mutagenicity tests was compared with the enzymatic activity in S9 fractions obtained from rats treated with either phenobarbital (NaPB), beta-naphthoflavone (betaNF) or combinations of both. Three pool systems prepared with different amounts of NaPB-induced S9 and betaNF-induced S9 were also analyzed for their activation capacities. Profiles of standard plate incorporation assays with Salmonella typhimurium TA98 increased with the amount of S9 fraction added for all drugs tested, except for 2AA, which showed a maximun of activity at low protein concentrations. According to these profiles, an optimal S9 protein content of 700-1000 microg/plate was estimated. For 2AAF and 3MTCL an S9 fraction obtained following a simultaneous treatment with NaPB (i.p.) and betaNF (oral gavage) (NaPB + betaNF) yielded the greatest response. This preparation was the only one which produced positive activation with 3MTCL as test drug. With the other test drugs all the S9 fractions were very active, including the NAPB + betaNF-induced S9. Both Phase I and Phase II cytochrome P450 enzymatic activities were enhanced in this S9 fraction. These results suggest that the simultaneous treatment (NaPB + betaNF) would be an adequate inducer for in vitro activation when used at 700-1000 microg protein/plate.