High air pollutions of PM2.5 concentrations have become a serious environmental problem in China during recent decades, causing significant influences on urban air quality and human health. In the study, we investigate the variations of the December PM2.5 in Eastern China and the possible causes during 2000–2020. The empirical orthogonal function (EOF) analysis is employed to reveal the dominant patterns of PM2.5 variability in Eastern China. The EOF1 shows a consistent variability in the whole of the Eastern China, which reflects a consistent emission pattern in Eastern China in past two decades. The EOF2 exhibits a North-South dipole pattern, which is closely tied to the changes of atmospheric circulations. The increase of PM2.5 in the North Eastern China is mainly related to the decrease of wind speed, the decrease of boundary layer height and the increase of inversion temperature, while the decrease of PM2.5 in the South Eastern China is affected by the increase of local precipitation. Two atmospheric wave trains are identified that affect the dipole distribution of PM2.5 in Eastern China. The southern one is affected by ENSO, and the northern one is jointly affected by ENSO, sea surface temperature of Labrador Sea and sea ice concentration near Kara Sea. Finally, we reconstructed a comprehensive atmospheric external forcing index based on these factors. We find that the comprehensive index can well reproduce the North-South dipole distribution of PM2.5 in Eastern China, indicating the plausible effects of the atmospheric external forcings and the prediction potential for the variations of PM2.5 in Eastern China.