In this paper we evaluate two approaches for estimating CO2 emission reduction from electricity savings: one based on average CO2 intensities of electricity generation and another that relies on marginal CO2 intensities. It is found that the average CO2 intensity approach has a significant shortcoming when it comes to scenario-based approaches for CO2 emission reduction. This shortcoming lies in the chicken-egg problem created, where larger future electricity savings are actually big enough to change the CO2 intensity in such a way that it cannot be used anymore to estimate the CO2 emission reduction. We show that in these cases the marginal approach is preferred. To correctly apply this approach, it is important to determine the CO2 intensity of the future power mix which will not be built in order to avoid under or overestimation of the CO2 savings calculated. We propose a seven-step approach which can be used in scenario-based potential studies as guidance for estimating the CO2 emission reductions from not only electricity savings but also renewable electricity and mitigation options that consume electricity such as electric cars and heat pumps. Using our approach would avoid a disconnection of the CO2 reduction potential with the underlying reference scenario. © 2013 Elsevier Ltd.