Most current research focusing on the health risk assessments of particulate polycyclic aromatic hydrocarbons (PAHs) have not analyzed the size distributions and human respiratory deposition rates. In the present study, size‐separated particulate matter (PM) was collected in the coastal area of Ningbo using an Anderson eight‐stage air sampler over a 1‐year period (2014–2015). The 16 US Environmental Protection Agency priority PAHs associated with PM were pretreated with rapid solvent extraction and analyzed by gas chromatography–mass spectrometry. The respiratory exposure assessment was determined using the multiple‐path particle dosimetry (MPPD) model. The results show that all PAHs exhibited bimodal distribution with one mode peak in accumulation mode (0.43–0.65 µm) and another mode peak in coarse mode (4.7–5.8 µm). In addition, a low coefficient of divergence of PAHs between PM2.1 and PM2.1−10 indicated a high spatial heterogeneity in source factor contribution and formation mechanism. The deposition fluxes (tracheobronchial + pulmonary) of PM were highest for children in the size range of 3.3 µm < particle diameter (Dp) < 9 µm, while for males and females the highest fluxes occurred in the size range of 1.1 µm < Dp < 2.1 µm. The depositions of coarse PM in children were significantly higher than those in adults. The benzo[a]pyrene equivalent (BaPeq) depositions of dibenz[a,h]anthracene ranged from 1.4e‐04 to 0.015 ng h−1, which were highest among the PAHs. The PAHs on particles with Dp >4.7 µm contributed approximately three times more to children than to males and females. Therefore, the toxicity of coarse PM to children needed attention. The incremental lifetime cancer risks (ILCR) for children, males, and females were estimated to be 2.92 × 10−7, 1.82 × 10−7, and 2.38 × 10−7, respectively, which were below the cancer risk guideline value (10−6). These ILCR values were much lower than the risks calculated without considering particle size distributions and respiratory depositions. The combination of the size‐segregated sampling technique and the MPPD model can effectively avoid the overestimation of human respiratory exposure. Environ Toxicol Chem 2024;00:1–14. © 2024 SETAC