By simulating vertical stratification data of PM2.5 with the third-generation air quality model CMAQ and high resolution relative humidity data with mesoscale meteorological model WRF, MODIS AOD data were revised by vertical and humidity correcting method, respectively. A linear regression model between revised AOD and PM2.5 was built, and the linear correlation coefficient was r = 0.77 (n = 57, P < 0.01). Based on this model, the average monthly concentrations of PM2.5 in 10 km resolution in January 2013 were firstly retrieved in the country, and the population exposure level was analyzed. The results showed that the areas where average monthly concentrations of PM2.5 was greater than 100 μg x m(-3) and 200 μg x m(-3) in January 2013 were 10.99% and 1.34% of the national terrestrial area, respectively, and the ratios of exposed population were as high as 45.01% and 6.31%, respectively.