The concentration of breathable particles (PM(10)) in urban areas has been associated with increases in morbidity and mortality of the exposed populations, therein the importance of this study. Organic compounds adsorbed to PM(10) are related to the increased risk to human health. Although some studies have shown the lack of correlation between specific mutagenic compounds in an organic complex mixture (OCM) and the mutagenic response in several bioassays, the same organic compounds selectively separated in less complex groups can show higher or lower mutagenic responses than in the OCM. In this study, we fractionated the OCM, from the PM(10) in four organic fractions of increasing polarity (F1-F4). The Salmonella bioassay with plate incorporation was applied for each one using TA98, with and without S9 (mammalian metabolic activation), and YG1021 (without S9) strains. The most polar fraction (F4) contained the greatest mass followed by F1 (non-polar), F2 and F3 (moderately polar). The concentrations of the OCM as well as the F4 were the only variables correlated with PM(10), atmospheric thermal inversions, fire-prone area, NO(2), SO(2), CO, rain and relative humidity. This indicated that polar organic compounds were originated in gas precursors formed during the atmospheric thermal inversions as well as the product of the incomplete combustion of vehicular exhausts and of burned vegetation. The percentages of the total PAH, and the individual PAH with molecular weight > or = 228 g mol(-1) (except retene) correlated with the percentages of indirect-acting mutagenicity in TA98+S9. The percentages of the total nitro-PAH and most of the analyzed individual nitro-PAH correlated with percentages of the direct-acting mutagenicity in both TA98-S9 and YG1021, the latter being more sensitive. In general, the highest mutagenic activity (indirect and direct) was found in F3 (moderately polar) and in F4 (polar). The non-polar fraction (F1) did not exhibit any kind of mutagenicity. In 77% of the cases, mutagenic activity was higher in the sum fractions with respect to their OCM. The combinations between F1, F2 and F4, with F3 under different or equal proportions suggested that mutagenicity reduction, in the combined matter of January (with TA98+S9 and YG1021) and of May (with YG1021), was due to concentrations of mutagens and non-mutagens in each fraction, and not to an antimutagenic effect. The organic compounds present in the non-polar fractions showed no antagonism, inhibition or reduction in the most mutagenic fractions in both indirect- and direct-acting mutagenicity, and the less polar organic compounds in F3 reduced mutagenicity in F4, in both months.