In order to elucidate the health effects of occupational exposure to traffic fumes, a few biomarkers of early genetic effect were investigated in Rome traffic policemen. One hundred and ninety healthy subjects engaged in traffic control (133 subjects) or in office work (57 subjects) participated the study. For all subjects, detailed information on smoking habits and other potential confounders were recorded by questionnaires. Average exposure of the study groups to benzene and other aromatic hydrocarbons was evaluated in a parallel exposure survey. All workers were genotyped for the following metabolic polymorphisms: CYP1A1 (m1, m2, and m4 variants), CYP2E1 (PstI and RsaI), NQO1 (Hinf1), GSTM1 and GSTT1 (null variants). In this paper, the results of the analysis of sister chromatid exchanges (SCE) in peripheral lymphocytes, and DNA damage by alkaline (pH 13) comet assay in mononuclear blood cells are reported. No statistically significant difference in the frequency of SCE or high frequency cells (HFC) was observed between traffic wardens and office workers (controls), despite the significantly higher exposure to benzene of the former (average group exposure 9.5 versus 3.8microg/m(3), 7h TWA). Conversely, both SCE per cell and HFC were highly significantly (P<0.001) increased in smokers compared to nonsmokers, showing a significant correlation (P<0.001) with the number of cigarettes per day. Multiple regression analyses of data, with metabolic polymorphisms, smoking habits, alcohol consumption, age, gender, and family history of cancer as independent variables, showed that smoking habits, and possibly the CYP2E1 variant genotypes, were the main factors explaining the variance of both SCE and HFC. Within smokers, an association of borderline significance between the CYP1A1 variant genotypes and increased SCE (P=0.050) and HFC (P=0.090) was found. This effect was mainly observed in light smokers (<15 cigarettes per day). The analysis of DNA damage by comet assay did not highlight any statistically significant difference between the exposed and control workers. Moreover, no significant model explaining tail moment variance was obtained by multiple regression analysis using the independent variables shown above. On the whole, these results indicate that exposure to moderate air pollution levels does not result in a detectable increase of genetic damage in blood cells. This evidence does not rule out any possibility of adverse effects, but strongly suggests that in urban residents life-style related factors, such as tobacco smoking, give the prevailing contribution to individual genotoxic burden.