Benzo(a)pyrene [B(a)P] air levels were measured in Florence (Italy) in the period 1992-2001. For the period 1999-2000 seven polycyclic aromatic hydrocarbons (PAH) (benzo(a)anthracene, crysene, benzo(a)pyrene (B(a)P), benzo(b)fluoranthene (B(b)F), benzo(k)fluoranthene, dibenzo(a,h)anthracene (DBA) and benzo(g,h,i)perylene (BGP)), were measured in the air in four different sites (one with heavy traffic (A), one in a park (B), one in a residential area (C) and one in a hill area (D)). B(a)P levels were elevated in 1992-1998 (maximum average value of winter months: 5.8 ng/ m3) but a decreasing trend was observed in the following years, probably due to improvement in vehicle emissions. The sum of PAH in the air in the period 1999-2000 was about one order of magnitude lower in the hill site (D) relative to the urban sites, and residential areas (B and C) had values 2.5-3 times lower compared to site A with a heavy traffic. PAH concentrations decreased in the warmer seasons of 2000 in all sites. A negative correlation was found between PAH levels and ozone. A positive correlation with carbon monoxide (CO) (r = 0.862, P < 0.001) and low B(a)P/BGP ratios, ranging from 0.44 to 0.51, indicated that vehicular traffic was the major PAH source in all monitored sites. Using B(a)P(TEF) values (toxic equivalency factors) for evaluating the biological activity of PAH, we found that the highest PAH contributors in terms of potential air carcinogenic activity were B(a)P and DBA. Therefore, in addition to B(a)P, DBA concentration should be considered in the evaluation of air quality in terms of PAH contamination.