To understand the impacts of different combinations of planetary boundary layer (PBL), short-wave (SW) and long-wave (LW) radiation schemes on the simulation results of meteorological variables and PM2.5 concentrations under extremely heavy pollution conditions, the Weather Research and Forecasting model with Chemistry (WRF-Chem) model was applied in Beijing to investigate a high PM2.5 pollution episode that occurred in January, 2013. Four PBL schemes, two SW schemes and three LW schemes with a total of 12 ensemble experiments were conducted in this study. The simulated meteorological variables including the temperature at 2 m (T2), the wind speed at 10 m (WS10) and the relative humidity (RH) were compared with their actual observations and the PM2.5 concentrations. A correlation analysis between the PM2.5 and T2, WS10 and RH values was also explored. The results indicated that there were no ideal scheme combinations that were most suitable for all meteorological variable simulations during this heavy pollution episode in Beijing. With the same emissions input, the simulation results of the WRF-Chem model that were configured with different physical parameterization schemes may vary significantly. As for the PM2.5 simulation, the combination of the YSU PBL, Goddard SW and GFDL LW schemes showed the greatest consistency with the observed values. Although the PBL schemes have the dominant impacts on the simulations of meteorological variables, the selection of LW and SW schemes is of the same importance.