Based on the Chinese provincial panel data collected from 2000 to 2014, the spatial correlation characteristics of PM_2.5 pollution are measured by using the exploratory spatial data analysis method, and then the spatial Durbin model is used to analyze the social and economic causes of PM_2.5 pollution. The results show that: (1) Both global and local spatial correlation characteristics of PM_2.5 pollution in China show positive spatial correlation, and the local spatial correlation characteristic mainly includes the two cluster areas: high-high type and low-low type. (2) The high-high cluster areas of PM_2.5 pollution are intensively distributed in the northern China plain and the Middle-Lower Yangtze plain while the low-low cluster areas are scattered throughout the Chinese northwest, southwest and northeast regions. (3) The increases of population density, industrialization level, economic development level and the number of motor vehicles will aggravate the degree of PM_2.5 pollution, while the improvement of urbanization level and technological progress can alleviate the degree of PM_2.5 pollution, but the role of environmental protection in lowering the degree of PM_2.5 pollution has not yet been revealed. (4) The increases of energy consumption can exacerbate the degree of PM_2.5 pollution in the local area, but its spatial spillover effects for the adjacent areas are not obvious.