Numerical simulations of long-term geological CO2 storage could be particularly useful in predicting the capacity, security, and other consequences of CO2 sequestration. The physics of this flow is essentially a multiphase multicomponent phenomenon in a strongly heterogeneous medium necessitating the need for a computationally prohibitive resolution to capture the various physical processes at all scales. Upscaling is an important step in these simulations when going from geostatistical models featuring strong heterogeneities to the simulation models which in practice are limited because of computational restrictions. Although many different upscaling techniques exist, including dynamic and steady-state methods, thorough analyses of their application to CO2 sequestration are not yet established. This review aims to highlight the recent developments in the application of upscaling to the modeling of long-term CO2 storage and provide insights into aspects that could prove valuable to numerical simulations.