Based on the real-time, on-line PM_2.5 monitoring data from 206 monitoring stations in Tangshan City, during three different periods, including heating period, heavy pollution period, and non-heating period, respectively, this research was to study the spatial distribution characteristics and spatial autocorrelation of air pollution so as to provide the basis to enhance air pollution control in Tangshan City, by using ArcGIS as the platform. All three periods had a certain spatial autocorrelation and the global spatial autocorrelation during heavy pollution period was the strongest. During heating and heavy pollution periods, the High-High cluster of PM_2.5 mainly occurred in the central areas of Tangshan, while the Low-Low cluster was mainly distributed in the northern mountain areas and a small portion in coastal areas. High-High cluster of non-heating period was mainly distributed in Fengrun District and Fengnan District, while Low-Low cluster occurred in the northern part of Zunhua. Through spatial interpolation, the PM_2.5 distribution of the whole city was simulated to result that the highest grid point value of PM_2.5 in Tangshan is 241μg/m³ during heavy pollution period, while the lowest value is only 37μg/m³ in non-heating period. Based on the characteristics of PM_2.5 concentration variation and spatial distribution, the 18 districts and counties of Tangshan were divided into 5 zones, and corresponding suggestions on PM_2.5 control measures for each zone were proposed.