The grey system theory is an interdisciplinary scientific area, the very method of analysis and prediction using small data sets and inadequate information to achieve an effective forecast of future trends optimally and generally. In this article, the mathematical principle, derivation and application are investigated. The data of sulfur dioxide emissions nationally and regionally from 1998 to 2012, are used to compare the performances and predict the emissions in the Twelfth Five-Year Plan period. The simulation results show that grey models have higher performances not only on model fitting but also on forecasting and some useful conclusions have been drew.