Abstract: Along with the construction and operation of the natural gas pipeline in the western and the eastern districts in China, the Shaanxi to Beijing Gas Pipeline, Zhongxian to Wuhan Gas Pipeline, the Sebei-Xi'ning-Lanzhou Pipeline and so on, the gas industry in China enters a new stage and the city natural gas utilization project are carrying on fiercely, therefore it is necessary to conduct the research on the city natural gas load characteristic of each kind of users as well as the existing load forecast method. In this paper, three kinds of forecasts methods which are commonly used at present are introduced: forecast method based on statistics theory; forecast method based on artificial intelligence and the other methods including grey model, the elasticity coefficient forecast and so on. The characteristics and application conditions of each method have been analyzed, so have the city natural gas load characteristics of each kind of users. Finally an optimization synthesis forecast model of natural gas load level is established which takes the historical data fitting degree, the development coordination uniformity and the model credibility as the judgment standard and it is suggested that the introduction of foreign advanced load forecast software should be speeded up to carry out the second development and application and develop the forecast software which adapt to the situation of domestic natural gas load forecast.