Abstract: Considering the spatial dependence of traffic accidents, factors that could affect the quantity of urban traffic accidents were studied. The study area was split into equally sized square cells and the quantity of urban traffic accidents and attributes was extractedin each cell. 11 variables such as relative location with the Second Ring Road,whether by fast road, number of schools and hospitals were takenas factors that could affect traffic accidents. OLS regression model, spatial lag model and spatial error model were established respectively and compared with each other. The study reveals that spatial error model which considers spatial dependence is superior to other models. More traffic accidents will happen in those cells that contain more hospitals or schools. Moreover, hospitals have greater influence on traffic accidents than schools have. In addition, those cells that contain stations have more traffic accidents because of locating in margin of city.