The influence of the longterm linear trends on the seasonal forecast skill of the surface air temperature in summer and winter over China is investigated using both observations and ensemble seasonal forecast output from the numerical model IAP AGCM （Atmospheric General Circulation Model from the Institute of Atmospheric Physics, Chinese Academy of Sciences）during the period from 1979 to 2004. The results show that the temperature during the 26 years show a warming trend over most of China while the warming trend in winter is more significant than that in summer. IAP AGCM ensemble forecasts cannot reasonably capture the characteristics of the longterm liner climate trends. Further study shows that the seasonal forecast skill of the surface air temperature over China can be significantly improved if the numerical model can perfectly capture the longterm linear trends in the forecasts.