In this study, a combination of monitoring data and the CAMx-PAST (comprehensive air quality model with particulate matter source apportionment technology) model was used to evaluate the meteorological effect of changes in the source apportionment results of PM2.5, SO42−, NO3−, NH4+ and their precursors (SO2, NO2, and NH3) in Beijing in January 2015, January 2016 and January 2017. The meteorological conditions in 2017 were more favorable than those in the other two years in terms of air pollution accumulation and formation, with a monthly high relative humidity of 72% and a low wind speed of 2.06 m s−1. PM2.5 pollution was more serious in 2017 than in other years, with an average concentration of 116 μg m−3. Both simulation and observation data demonstrated that nitrate accounted for the largest proportion of the total secondary inorganic aerosol (SO42−, NO3− and NH4+). In the winter in Beijing, residential areas were a major source of pollution and contributed to 51%~54% of the total pollution. Meteorology has a great influence on the regional transportation of pollution. The transport ratios of PM2.5 were 22%, 26% and 33% in 2015, 2016 and 2017, respectively. Regional transport has a significant effect on both SO42− and NO3− concentration. However, primary pollution is mainly affected by local pollution sources. Local residential emissions have the greatest impact on SO2, contributing 85%, 82% and 83% of the total concentration. The local emissions contributed 88%, 84% and 84% of the total NO2 concentration.