In this research, the improved Terrestrial Ecosystem Regional (TECO-R) model was adapted to steppe ecosystems and then utilized to simulate the soil organic carbon pool in the period from 1989 to 2011 (excluding 1994, 2002, 2009, and 2010) for a typical steppe in Xilingol League of Inner Mongolia in China. The improved TECO-R model is an ecological model in combination of remote sensing data, which allows the spatial scale for the analysis of soil organic carbon which is not limited to vegetation or soil type. The spatial and temporal resolution advantages of remote sensing image can be well utilized in this model. The results indicate that in addition to an accurate simulation of the soil carbon pool of a steppe ecosystem, the vegetation aboveground carbon pool, grazing intensity of herbivores, mowing coefficient, litter carbon pool, root carbon pools of different vegetation layers, root-shoot ratio, actual residence time of different carbon pools, and allocation coefficients of different carbon pools in corresponding years are also available from the TECO-R model. Some of the above data are difficult to obtain through macro-observation but can be simulated with the TECO-R model by combining the model with input data; this is very important for a correct understanding of the feedback relationships between the steppe ecosystem’s carbon cycle and climate change (e.g., global warming) and human activities such as grazing.