We consider the socially optimal use of solar geoengineering to manage climate change and its implications for carbon emissions abatement policy. We show that solar geoengineering is a substitute for emissions abatement; optimal policy includes less abatement, by up to eight percentage points, and has a lower carbon price, by up to fifteen percent, than recommended by models that ignore solar geoengineering. However, it is an imperfect substitute, since it reduces temperature without reducing atmospheric or ocean carbon concentrations. Carbon concentrations are higher but temperature is lower when allowing for solar geoengineering. Ignoring geoengineering in climate models can lead to welfare losses of up to 4 percent of GDP. Uncertainty over climate sensitivity leads to more abatement and solar geoengineering, while uncertainty over solar geoengineering damages leads to less geoengineering.