The desertified land is a complex earth surface system that induced by both natural factors and human activities.It is meaningful to develop a desertification simulating model coupling with natural and human factors for understanding the desertification progress,forecasting desertification trend and controlling desertification expansion.In this paper,Ordos was selected as the typical research region,and system dynamic method was used to construct the impacts and feedbacks among climate,soil water,people,economy,pasturage, land use and vegetation Net Primary Production (NPP);based on defining the threshold value of NPP for different desertification grades,the system dynamic model coupling with natural and human factors for desertification simulation was developed.The comparison between simulated NPP and actual NPP derived from MODIS images showed that the model developed in this study was reliable.The simulated results of vegetation NPP and desertification status from2011 to 2030 under different scenarios showed that rainfall was the dominant climatic factors for desertification dynamics in Ordos;however,50%increase of rainfall could not change the desertification status without the change of human activities in future.Comparing with grain for green and afforestation policies,the grazing prohibition and rest policy would be more beneficial for vegetation rehabilitation,but the cost would also be higher.The solely implementation of any pasturage or land use policy would not change the status of desertification without climate change;however,abolishing any of them would lead the status of desertification develop to severe and very severe in future.For local government,it is necessary to choose different combination of policies according to finance status to facilitate desertification reversion.