Abstract: Based on NASA’s global annual average PM2.5 grids, by vectoring the administrative boundaries of African countries, calculating the average concentration of atmospheric PM2.5 of all countries and building the spatial database, using gravity model, ESDA model and GIS spatial statistical analysis method, we explore the space-time pattern evolution of the atmospheric PM2.5 pollution concentration of 52 main African countries from 2001 to 2010. According to their time series features, we classify them into 8 types. The results show that: (1) The concentration of atmospheric PM2.5 pollution in Africa generally shows the spatial characteristics of "middle high, north and south low; west high, east low" during 2001 to 2010. High concentration areas such as Nigeria, Congo and Cameroon, are mainly concentrated in the vicinity of the Gulf of Guinea in West Africa, and low concentration areas are widely distributed in North Africa, Southern Africa and coastal areas or islands of Southeast Africa. (2) Using spatial autocorrelation analysis method based on ESDA Model, we find that "High-High" type areas of PM2.5 concentrations are mainly distributed in the major countries near Gulf of Guinea, and that "Low-Low" type areas of PM2.5 concentrations are concentrated in the countries of Southeast Africa such as South Africa, Mozambique and Madagascar. (3) The average annual PM2.5 concentrations in Africa decreased obviously from 2001 to 2010. Average annual PM2.5 pollution concentrations of 32 counties (accounted for 61.54% of total) in 2010 were lower than in 2001. (4) Due to theirs high population density and dependence on the oil industry, countries near Gulf of Guinea are the areas with the highest PM2.5 concentration in Africa. On the contrary, the average PM2.5 concentrations of Southeast Africa is the lowest, thanks to their natural environment and low-pollution pillar industries.