Based on the observed monthly surface temperature and daily precipitation of 550 stations in China from 1961 to 2000, and the surface temperature and precipitation simulated by 13 IPCC AR4(the fourth assessment report of the intergovernmental panel on climate change) models and their multi-model ensemble(MME) mean, the simulation abilities of the 13 models are evaluated. Results show that the IPCC AR4 models certainly have a capability of modeling the surface temperature and precipitation changes in China. The annual variation and spatial distribution of surface temperature are well simulated, but the simulations are systematically smaller than the observations in the simulated region with the well simulation in the eastern China than that in the midwestern China. The annual variation and spatial distribution of China precipitation can be well simulated by the most models, but the simulations have the large differences among the different simulated regions. The most models can well simulate the northward process of summer monsoon rain belt in eastern China, but the simulated rain belt lies poleward of its observed position. The updated IPCC AR4 models can simulate the linear trend of surface temperature, but they have poor performances in simulating the interannual variability of surface temperature and precipitation. Comparisons of many evaluation indexes indicate that the surface temperature simulation of the multi-model ensemble(MME) is the best, while the precipitation simulated by UKMO-HadCM3 is the best among the 13 models and MME.