Molecular and carbon isotopic compositions of atmospheric polycyclic aromatic hydrocarbons (PAHs) were measured in sites at 1 m (S1), 10 m (S2) and approximately 200 m (S3) distant from roads in Tokyo, Japan. Total suspended particles (TSP) concentrations at S1 was approximately three times higher than S2 and S3, while the concentration of atmospheric PAHs was almost the same (76-166 µg g -1 -TSP) at S1, S2 and S3. Compound-specific δ 13 C of atmospheric PAHs ranged from -23.3 to -24.5‰ at S1, from -24.1 to -25.6‰ at S2, and from -25.2 to -27.0‰ at S3. The concentration of some PAHs and the weighted averages of δ 13 C of some selected PAHs differed significantly among the three sites. The PAH concentration ratios indicating automotive contribution were S1 > S2 > S3, while the ratios indicating non-vehicle contribution were S3 > S2 > S1. These differences in molecular and isotopic signatures among the three sites are not caused by photolysis, instead they are possibly due to differences in the sources of PAHs.A quantitative approach to identify compound-specific sources is to use the concentration and δ 13 C of individual PAHs at each site. The estimated average δ 13 C values of individual PAHs derived from automobiles, which was calculated using a mass balance model, ranged from -19.5 to -23.6‰. Taking into account the results of this study and those of previous studies, -26 to -28‰ was adopted as the δ 13 C value of PAHs derived from non-vehicle sources. We estimated that the automotive contribution to the concentration of individual PAHs ranged from 33 to 88% at S1, from 11 to 74% at S2, and from 0 to 52% at S3. We conclude that compound-specific δ 13 C measurement is a powerful and useful tool to identify such compound-specific sources of atmospheric PAHs.