Abstract: In order to predict the actual carrying capacity and theoretical carrying capacity, and to provide the theoretical basis for protecting the ecological environment of grassland and the sustainable development of grassland-animal husbandry, the paper analyzed the real carrying capacity and theoretical carrying capacity of grassland in 4 countries of Gannan in China from 1998 to 2009, based on grey theory and the Markov process, and built a optimized and improved Grey-Markov forecasting model for grassland carrying capacity. The model is equivalent dimensions additional grey model and Markov residual forecasting model. The model was tested by actual stocking capacity in Hezuo country. The prediction by the model had no significant difference with reality (p<0.01, error less than 7.06%). This showed the availability of the model for predicting the carrying capacity of grassland. Furthermore, actual carrying capacity and theoretical carrying capacity in each county of Gannan from 2010 to 2014 were predicted. This study can provide a reference for the relevant departments to determine the reasonable carrying capacities and the grassland management.