Abstract: The grassland net primary production(NPP) is an important research issue in the global climate change and terrestrial ecosystem. The grassland NPP simulation methods have been developed from the sampling observation and statistical models to process models. The sampling observation NPP data can be used as baseline for evaluating the estimation of NPP results of statistical models and process models. Several NPP parametric models and process models are compared in this paper. These models range in complexity from regressions between climatic variables and NPP to quasi-mechanistic models that simulate the biophysical and ecophysiological processes. The statistical models account for the relationships between NPP and climatic variables (i.e. temperature, precipitation, radiation) or directly calculate NPP using vegetation indices(VIs) derived from remote sensing information. The process models for estimating NPP simulate a series of plant ecophysiological and biophysical processes on the basis of plant physical and physiological principles, the major processes are photosynthesis, growth and maintenance respiration, evapotranspiration, uptake and release of nitrogen，allocation of photosynthesis to the various parts of the plants, decomposition, and phenological development, etc. The process models based on remote sensing data can simulate grassland NPP timely, dynamically and macroscopically, and determine the NPP on large spatio-temporal scales by using input parameters, i.e., the landcover, LAI, surface albedo, surface temperature, soil moisture condition, derived from remote sensing data. In the end, the main problem and development trend of grassland NPP study by using remote sensing data were discussed. It indicates that process models based on remote sensing data can improve the temporal-spatial precision of NPP simulating result with the development of remote sensing techniques.