Abstract: Using the T639 model forecast products and air temperature observations at 83 national weather stations in Heilongjiang province, we selected forecast factors using an optimal selection method, and established the Model Output Statistics (MOS) prediction equations for daily maximum air temperature (TMAX) and daily minimum air temperature (TMIN) using a multiple regression method.In addition, we comparatively analyzed and validated the forecast performance of TMAX and TMIN from the MOS, the guide forecasts of the Central Meteorological Observatory (SCMOS), and three air temperature forecast products from the T639 model, and examined the consistency of spatiotemporal distribution between the predicted and observed air temperature using the Empirical Orthogonal Function (EOF) method.The results showed that the MOS and SCMOC perform better in predicting the spatiotemporal distribution of air temperature, while the T639 model performs relatively poorer.The values of 2℃ forecast accurate rate (TT₂) for TMAX and TMIN from the MOS and SCMOC are mostly higher than those using the T639 model, and the TT₂ values for TMAX/TMIN from the MOS are higher/lower than those from SCMOC.MOS can improve the air temperature forecast from the T639 model, especially for the TMIN forecast in winter.There is a negative correlation between the improvement of MOS relative to the T639 forecast and the performance of the T639 forecast.The MOS's improvement is better over mountain areas with low TT₂ predicted by the T639 model than that over plain areas.In spring and summer, the MOS's improvement is better for the TMAX with low TT₂ than the TMIN, while in winter, the MOS improvement is better for the TMIN with low TT₂ than the TMAX.This MOS air temperature forecast method has a good forecast capability and can be applied to the interpretation and application of other numerical model products.The SCMOC can be used in the TMIN forecast in Heilongjiang province considering its good forecast performance; the TMIN parameter is usually difficult to forecast in Heilongjiang province.