Decreasing the greenhouse gas (GHG) emissions from the residential sector is critical to the low-carbon transition. Applying information and communication technologies to power systems makes it possible to reduce GHG emissions in the residential sector, for example through the development of smart homes. Smart homes are more energy efficient and thus, they may be prone to the rebound effect (RE), (i.e., an increase in demand following the introduction of more efficient technology). Moreover, because the electricity's environmental impacts, cost and demand all vary over time, the potential for RE may also fluctuate. Accounting for these temporal aspects could therefore provide more insights into how and why potential RE may occur in smart homes, especially with regard to households' behaviours. In this study, an agent-based model is used to simulate standard and smart home electricity consumption. Life cycle assessment and environmentally extended input-output tables are used to calculate the households' electricity consumption and RE GHG emissions during the simulations. Results show that, while indirect RE in smart homes is low (about 5% in the simulations), the choice of metric used for smart electricity management is key to maximize the GHG emissions reductions of smart homes. When smart homes perform load shifting based on an economic rather than environmental signal, RE increases by almost five-fold. Moreover, certain periods, such as weekdays or the winter season, lead to more significant RE. Thus, considering factors that decrease RE could enable smart homes to reach their full potential contribution to sustainability. © 2020 Elsevier Ltd