Climate change has prompted scientists to search for possible ways of reducing CO 2 emissions or even capturing it from the atmosphere. Catalytic reduction of CO 2 into value-added chemicals has been put forward as a viable strategy. While thermocatalytic routes of producing CO, methanol, methane, and higher hydrocarbons from CO 2 have been the focus of considerable research efforts, photocatalytic conversion is an emerging approach. Photoactivation of CO 2 has potential as a greener process because it could be carried out at lower temperatures and pressures, decreasing the energy consumption. The recent advent of available computational power and tools has made it possible to study the reaction in silico for catalyst prediction and mechanism elucidation. In this Review, we thus focus on ab initio research of photocatalytic CO 2 reduction and comparison with experiments. The most commonly used materials are variously doped TiO 2 , g-C 3 N 4 , and perovskites, which have favorable optical properties on their own. Their efficiency is mostly governed by the band gap, charge separation, and charge transfer. Their characteristics can be improved, and the catalysts can be tailored for a specific use by doping, introduction of defects (such as oxygen vacancies or geometric effects), cocatalysts, or using Z-scheme catalysts. Most theoretical studies focus on the calculations of conduction and valence bands, band gap, and adsorption energies. Some studies try to describe the reaction mechanism. Few studies go beyond the hybrid-functional DFT approach and try to explicitly model the effect of electron excitation. While a theoretical description of the excited states is possible with post-DFT methods, it has yet to be applied to the problem of CO 2 photoreduction on a larger scale. We conclude the Review with an outlook on how the current state-of-the-art can be used for improving the existing catalysts.