Since the view that energy efficiency improvement would increase rather than reduce energy consumption, so-called rebound effect, was put forward and has been the focus that is encouraging a fierce debate within mainstream economics. This paper tries to introduce an improved estimation model by efficiency share decomposed from energy intensity, based on IPAT equation to estimate the rebound effect of three major regions in Guangdong, together with redistricting Guangdong applying hierarchical clustering analysis. The result shows the average rebound effect of low-intensity high-yield area, high-intensity low-yield area, and low-intensity low-yield area, is 17.91%, 10.82% and 8.10%, correspondingly. While the empirical study provides support of existence of rebound effect, it does suggest we should not only rely on improving energy efficiency, but also using market instruments such as resources taxes, to reinforce the energy savings.