Aerosol filter samples collected at a major urban traffic junction (LKP) and at a suburban residential location (IWC) in the metropolitan area of Munich (Germany) throughout the years 2001 and 2002 have been analyzed for 12 of the 16 EPA priority polycyclic aromatic hydrocarbon (PAH) pollutants by liquid chromatography with fluorescence detection. The mean mass concentration of the sum of all investigated PAH in the sampled air at LKP (1.9-5.0 ng m(-3)) was roughly two times higher than at IWC (0.8-2.9 ng m(-3)), and at both locations it was about 2-3 times higher in winter (heating season) than in summer and spring or autumn. Comparisons with earlier measurement campaigns indicate a steep decrease of PAH abundance by almost an order of magnitude from 1980 to 1993 and a much slower decrease since then. Distinctly different seasonal trends and short-term fluctuations have been observed for semivolatile 3- and 4-ring PAH and for particle-bound 5- and 6-ring PAH. Based on systematic correlation analyses with a wide range of air quality parameters, most of the differences can be attributed to not only varying emissions but also chemical reactions with atmospheric oxidants which were found to play an important role. The results of denuder experiments prove that substantial degradation of the particularly toxic tracer benzo[a]pyrene and of the other investigated 5- and 6-ring PAH can occur during filter sampling and on airborne particles (formation of oxygenated and nitrated derivatives). Filter reaction artifacts are shown to lead to an underestimation of the actual PAH content of urban air particulate matter by up to 100% of the measurement value or more, with a near-linear dependence on ozone volume mixing ratio. The role and applicability of ozone as a tracer of atmospheric oxidizing capacity for particle-bound PAH is discussed and confirmed by comparison with earlier investigations and by complementary laboratory experiments (reaction kinetics and product studies).