Based on the daytoday sounding data at Xuzhou, Nanjing, and Sheyang in July and August from 2002 to 2006, the models to forecast maximum temperature are established for Xuzhou, Nanjing, and Sheyang using the stepwise regression statistic method and through analyzing and electing some factors influencing temperature. The regression results and forecasts of the model are analyzed thoroughly, and the rationality and credibility are discussed. In combination with dynamical numerical forecast results of ECMWF, using the Gaussian weight interpolation to insert grid data of the forecast field over Jiangsu Province, the maximum temperature quantitative forecasting for Jiansu Province through the PP method is performed.