Abstract: Inter-annual and inter-decadal variability are two kinds of different timescale variability existing at the same time in climate system found in previous studies. Affected by the global warming, the inter-decadal signal of climate change becomes more and more significant. The next 10 to 30 years of climate change, namely inter-decadal time scales climate change and their impacts on the global environment, society and economic development, draw more and more attention. Climate change features of inter-decadal scale become one of the most important content of the IPCC AR5. The 10 to 30 years' timescale of inter-decadal forecast experiment which is listed as one of the main experiment content has joined the 5th Coupled Model Inter-comparison Project (CMIP5). More in-depth research will be carried out on predictability of inter-decadal timescale.The air temperature data of 541 stations in China from 1960 to 2010 as well as the CMIP5 historical and decadal experiment results of Beijing Climate Center Climate System Model (BCC_CSM1.1) are utilized to evaluate the simulation ability of the model. The model results are interpolated to the corresponding latitude and longitude of 541 stations use bilinear interpolation method. Whether the pattern of regional prediction ability could improve by the decadal experiment of BCC_CSM1.1 which initialed the SST (sea surface temperature) is discussed. Bias corrections to the decadal experiment results are done and the preliminary projection of the changes of the air temperature of China for the next 10—20 years is presented. Results show that both historical and decadal experiments can capture the warming trend in accordance with the observations, but the warming tendency of the experiments are less significant than those of observations. Results of historical experiments are slightly better than those of decadal experiments of the model. On the inter-decadal timescales, simulations in the eastern part of China are better than those in the western part of China. On the inter-annual timescales, the high prediction skills are located in the southwestern and eastern parts of northwest region, and southwest of China. Distributions of temperature in China are well simulated in both of historical and decadal experiments, such as the spatial correlation coefficients of 0.9 or above. After bias correction, results of decadal experiments are much better. By the corrected result of decadal experiments, the result of temperature spatial distribution simulation is better. The model projects that the rising rate of the mean temperature of China will be 0.48℃/10 a during 2011—2030, which is more significant than the warming rate of 0.27℃/10 a during 1960—2010 on the basis of observations. And the forecast results of the model show that the air temperature of China during 2001—2010 grows more slowly and fluctuate less compared with the period of 2011—2030.