Abstract: In order to fully grasp the change rule of temperature in greenhouses, to forecast the extreme temperature within greenhouse and to improve the service support capability of agricultural meteorological, the author used the inside the greenhouse temperature contrast hour observation data in Chifeng of eastern Inner Mongolia in 2012. The author analyzed the diurnal variation of the temperature within greenhouse, typical daily variation, the highest temperature and minimum temperature and the number of occurrences of probability, and analyzed the relationship between the temperature inside and outside the greenhouse, and then used the method of regression analysis, respectively, established the greenhouse temperature prediction model in sunny day. Results showed that: the daily variation of the temperature difference of different season, the highest and lowest temperatures of each season there was not the same time and the temperature of the regression analysis method to build the prediction model had better effect on the lowest temperature forecast.