Much attention is given to the distributions, sources, and health risks of atmospheric polycyclic aromatic hydrocarbons (PAHs) in cities. In this study, 62 total suspended particle samples were collected from April 2013 to March 2014 in Lhasa city. Positive matrix factorization (PMF) was applied to investigate the source apportionment of 15 priority PAHs, the lifetime carcinogenic risk (LCR) levels of which were assessed. The annual average particle phase PAH concentration was 43.9 ± 60.4 ng m–3. Evident seasonal variations of PAHs were observed, with the highest concentration observed in winter, followed by autumn, spring, and summer in descending order. Four- and five-ring PAHs accounted for the predominant proportion (63.3%–84.4%) throughout the year. Correspondingly, gas phase PAHs showed opposite variations with the highest and lowest concentrations observed in summer and winter, respectively. And three-ring PAHs especially Ace, Acel, and Flu contributed most. Compositions of particle phase PAHs varied seasonally, with four-ring PAHs contributing more in winter than in summer and five-ring PAHs exhibiting the opposing trend, thereby reflecting the variety of emission sources. PMF analysis showed that biomass combustion (48.4%) and vehicle emissions (27.9%) were the main two PAH sources, followed by coal combustion and air–surface exchange. These results were consistent with those of diagnostic molecular ratios. The benzo(a)pyrene equivalent (BaPeq) concentration of particle phase PAHs ranged between 1.48 and 24.5 ng m–3, all of which were higher than the new limit value in China (1 ng m–3). The average BaPeq of gas phase PAHs were 6.43 ± 4.15 ng m–3, which were similar with those of particle phase PAHs. The LCR of total PAHs (9.08 × 10–6) was one time higher than that of particle phase, however, a little lower than the acceptable level, thereby indicating that atmospheric PAHs in Lhasa pose no or little carcinogenic risk to locals.