A large accumulation of carbon dioxide and other greenhouse gases have caused great concern around the world. A great deal of general literature focus on the impact factors of CO 2 emissions at the national, regional and city levels. However, there is little specific guidance on regional difference in CO 2 emissions. In this paper, 30 provincial-level administrative units of China are divided into three different levels of economic development regions according to the GDP per capita from 1997 to 2012. A STIRPAT (Stochastic Impacts by Regression on Population, Affluence and Technology) model is used to examine the impact factors on energy-related CO 2 emissions, including population, economic level, technology level, urbanization level, industrialization level and foreign trade degree. The results indicate that the effect of energy intensity is the greatest in highly developed region. Nevertheless, the impact of urbanization, industry structure and foreign trade degree in under developed region is higher than the other two regions. Population and GDP per capita have greater effect on carbon emissions in developing region than the others. Finally, differentiated measures for CO 2 reductions should be adopted according to local conditions of different regions.