A deep understanding of the environmental behavior of polycyclic aromatic hydrocarbons (PAHs) in combined sewer overflows (CSOs) is essential for evaluating and mitigating PAHs in the urban environment. This study characterized occurrence of 16 PAHs in CSOs, the source of which were analyzed by the positive matrix factorization (PMF) models and the self-organizing map (SOM). The adverse effects of 16 PAHs in CSOs on the microbial diversity and functional gene of the receiving water sediments were analyzed through Illumina MiSeq sequencing and PICRUSt2 platform. The results showed that due to the relatively low vapor pressure and easy adsorption on particles, the proportion of tetracyclic PAHs was the highest among 16 PAHs from the CSOs with an average value of 42.5%. The first flush effect of PAHs had been found in each rainfall event. However, there was no significant correlation between the first flush effect and the rainfall characteristics. In terms of source apportionment, the SOM and PMF models identified vehicular emissions as the predominant source of particulate PAHs from the CSOs, which accounted for ≥54.8% of the total PAHs. ∑₁₆PAHs levels in the river sediment were directly affected by two gravity pipe outlets and three pumping station outlets, which were significantly higher than the control points. Besides, PAHs with different benzene rings had a significant correlation with specific microbial populations. The PAH related genes were also specifically affected by the CSO events. Therefore, through the observation of gene function, we could better find the specific impact of CSOs on the river's ecological environment.