Abstract: The construction of estimation model with remote sensing for grassland net primary productivity (NPP) is one of the effective ways to realize grassland NPP estimation in large areas. This paper analyzed the relationships between NPP and normalized difference vegetation index (NDVI) based on the moderate-resolution imaging spectroradiometer (MODIS) data and field measurement data. The estimation model of NPP in the southern grassy mountains and slopes was constructed with NDVI as the independent variable and hydrothermal conditions as regulatory factors, and the model was validated by independent observed data in different years. There were five relevant types between the grassland NPP and NDVI, and the correlations all reached a very significant level. There were good correlation and consistency between the simulated and observed NPP, and R² were 0.902 2, 0.826 6, 0.871 2, 0.887 7, 0.875 5, respectively, all achieved a very significant level too. The RMSE and RRMSE between observed and simulated NPP were smaller. It indicated that the model was reliable, and the results of above provided an effective method for the estimation of grassland NPP and the resource management in southern china.