After the implementation of the Coal Replacing Project (CRP) in the northern parts of China in 2017, its effect on PM2.5 composition is still unclear. In the study, water-soluble ionic components (WSICs) and stable sulfur isotope ratios (δ34S) of SO42− in PM2.5 collected during the domestic heating period before and after the implementation of CRP in Tianjin were analyzed. Results showed that the average concentrations of both PM2.5 and WSICs have dropped dramatically after the CRP, especially for the SO42− (by approximately 57–60%). After the CRP, the range of δ34Ssulfate was significantly narrowed to 4.1–7.5‰ in January 2018 and 1.4–6.1‰ in January 2019, which suggested that the sulfur source was becoming simple. It was interesting that the δ34Ssulfate value in the pollution period before the CRP was higher than that in the clean period, whereas it showed the opposite tendency after the CRP, which implied that the contribution of sea salt was high during the pollution period before the CRP. The MIXSIAR model calculated that the contributions of the transition-metal ion (TMI) oxidation and NO2 oxidation pathways in the three sampling stages were higher than those of the OH radical oxidation and H2O2/O3 oxidation pathways, indicating that the formation pathway of sulfate was mainly dominated by heterogeneous oxidation. Before the CRP, the NO2 oxidation pathway was the dominant sulfate oxidation pathway during a haze episode, and the TMI oxidation pathway dominated the formation of sulfates after the CRP.