Energy substitution is an effective way to deal with CO
2 emissions. Considering the complexity of energy substitution’s effect on carbon emissions, this study designs a new three–tier structural decomposition analysis approach to decompose the factors affecting carbon emissions. We apply this approach using Chinese input–output data, energy consumption data, and carbon emission data for 22 consecutive years from 1997 to 2018. The first–tier decomposition results indicate that the energy intensity effect is the main driver of carbon emission reduction. The second–tier continues to decompose the energy intensity effect and shows that the energy substitution effect gradually becomes stronger, but it is unstable. The third–tier decomposes the energy substitution effect into three dimensions based on energy type, industry energy consumption, and the mode of influence. We find that substituting other energy for row coal and coal products can effectively reduce carbon emissions, and that non–high energy consumption sectors can better reduce emissions than high energy consumption ones. This study sheds light on the deeper patterns of energy substitution, thus providing timely insights for policymakers working on energy transition and carbon emission reduction.