Abstract: In order to reveal the variation rule of temperature inside and outside the solar greenhouse, using the meteorological data inside and outside the solar greenhouse in Datong County of Qinghai Province, the authors established the maximum and minimum temperature forecasting model based on the stepwise regression method. The results showed that: (1) the daily temperature variation in the greenhouse had a‘single peak’type in different weather types, indicating that the variation amplitude was sunny＞cloudy＞overcast days; (2) the daily maximum and minimum temperature inside the greenhouse was significantly higher than that outside; (3) the correlation analysis manifested that the correlation between the daily maximum and minimum temperature and the percentage of sunshine inside and outside the greenhouse was the highest, while the correlation between the daily maximum and minimum temperature and the 10 cm ground temperature was the lowest; (4) the model could well forecast the maximum and minimum temperature inside the greenhouse in different weather types, and the simulation result in cloudy and overcast days was better than that in sunny days. The result provided a scientific basis for avoiding the low temperature (freezing) damage in agricultural production in alpine cold regions, and mastering the suitable sowing time of winter vegetables.