Abstract: Grassland soils are considered to have great potential for carbon sequestration. It is an essential part and also one of the biggest challenges to detect the soil organic carbon storages of global grassland ecosystem for regulating the global carbon cycle. This paper reviewed the research progress on soil organic carbon stocks of grassland ecosystem, and analyzed the main technical features of the existing models for simulating soil organic carbon. The existing technical weakness of the models, especially involving data accuracy, model structure and function formulization, are inquired in detail with relevant suggestions for improvement provided. The integrated use of sample inventory, remote sensing analysis, and simulating models would be the fundamental way to solve this problem. Finally, a framework of the Classification Indices-based Model on grassland soil organic carbon was presented based on the comprehensive sequential classification system of grassland (CSCS). Coupling the sample inventory and the Classification Indices-based Model with the high spatial and temporal resolution characteristic of remote sensing, the accuracy of the estimating grassland soil organic carbon will be improved greatly among the different grassland classes, ecological or climatic zones. There is no absolute conflict between the capacity of grassland soil carbon sink and grazing utilization, and precise estimation of grassland soil organic carbon is a premise of the grazing modern transformation.