The temperature and precipitation observation data in Haihe river basin during 1961-2010 are used to evaluate the simulation performance of 20 IPCC AR4 Climate Models and Multi-Model ensemble.The results show that all the climate models underestimate surface air temperature,and spatial correlation coefficients reached 0.7.Most climate models overestimate precipitation in the Haihe river basin.Only a few models(MIUB_ECHO_G model and Multi-Model ensemble) show roughly consistent with the observation data.The projection results show that the temperature would increase,especially under SRES-A1B,the annual temperature increase is higher in Haihe river basin than in other parts of China projected by all climate models.Future precipitation would rise too,and significant precipitation increase is projected by all models under SRES-A1B and SRES-B1.The spring precipitation would increase significantly under SRES-A2.Catastrophe point of MIUB_ECHO_G model appears in 2013 under SRES-A1B,when precipitation would increase drastically.While under SRES-A2,catastrophe point of MIUB_ECHO_G model and Multi-Model ensemble appear in 2031 and 2001,respectively.