This study focuses on the source apportionments of polycyclic aromatic hydrocarbons (PAHs) in road dust (RD) with four size fractions through three receptor models of principal component analysis with multiple linear regression (PCA-MLR), positive matrix factorization (PMF) and Unmix. The concentrations of total PAHs range from 0.45 to 2.03μgg−1. Results show that the concentrations of PAHs increased with a decreasing size fraction. Similar potential sources to PAHs in RD were extracted by three models with a little difference in numbers and percent load contributions of each identified sources. The overall proportion of the identified sources were ranked as vehicular emission>coke oven>surface pavement>others in each size fractions. In terms of risk assessment, the mean values of incremental lifetime cancer risk (ILCR) of the total cancer risk of PAHs in RD were lower than the baseline value of an acceptable risk. However, PAHs in smaller size fraction prone to have a higher adverse effect on children via ingestion. Furthermore, the ecological risk assessment of hazard quotients and mean hazard quotients indicated that PAHs in RD had a 9% probability of being toxic to the benthic organisms and aquatic environment. Display omitted •PAHs contents in road dust decreased with an increasing size fraction.•Model comparison between PCA-MLR, PMF and Unmix.•PCA-MLR identified a lower number of PAHs sources.•The differentiation of vehicular emissions by Unmix was dataset dependent.•PMF could give a more complete source apportionment.•Traffic-related sources were primary contributors to PAHs in road dust.