The main aim of this study was to compare the genotoxic potential of organic extracts from urban air particles collected in three different sampling periods in the center of Prague (Czech Republic). For this purpose, we analyzed the DNA adduct forming activity of extractable organic matter (EOM) from urban air particles <10 microm (PM10) in the human hepatoma cell line HepG2. DNA adducts were analyzed by (32)P-postlabelling with nuclease P1 enrichment. PM10 concentrations were 36.9 microg/m(3), 62.6mug/m(3) and 39.0 microg/m(3), in summer 2000, winter 2001 and winter 2005, respectively. The corresponding EOM contents were 5.0 microg/m(3) (13.9% of PM10), 14.9 microg/m(3) (23.8%) and 6.7 microg/m(3) (17.2%). The total DNA adduct levels induced by 10 microg EOM/ml were 4.7, 19.5 and 37.2 adducts/10(8) nucleotides in summer 2000, winter 2001 and winter 2005, respectively. However, when the EOM quantities per cubic meter of air were taken into consideration, the summer sample exhibited a 10-fold lower genotoxicity than did those of winter, while the difference between the winter samples was not significant: 23.4 in summer 2000, 291 in winter 2001 and 249 in winter 2005 (in relative units). Although the PM10 concentration in air and the EOM content in particles in winter 2005 were significantly lower than in winter 2001, the genotoxic potential of the ambient air in these samples was almost equal. There were significant positive correlations between the B[a]P and c-PAH content in EOM from various sampling periods and the total DNA adduct levels detected in the EOM-treated samples. These findings support the hypothesis that the B[a]P and c-PAH content in EOM is the most important factor that determines its genotoxic potential. Thus, estimating the genotoxic potential of the ambient air and predicting health risk should be based mainly on the c-PAH concentration and the biological activity of the extracts, while the mass of particles and the EOM content do not seem to be crucial determinants of ambient air genotoxicity.