The paper is to identify the factors that influence the PM2.5 value. Considering that components of PM2.5 are complex, I simplify the problem by neglecting unnecessary factors. First, I build a hierarchical structure model, consisting of five main factors that can influence PM2.5 value: Vehicle exhaust, Fossil fuel burning, Industry, Flowing dust and other factors. The criterion and plan layer consist of elements of PM2.5 and factors, respectively. The goal, PM 2.5 value, is the objective layer. Second, comparative matrix pairs are constructed. Third, to check whether models built are reasonable, I make consistency test. Based on the consistency ratio (CR) that is calculated below 0.1, I believe that the totality passes the consistency check. Finally, based on their weights five main factors are sorted.