When changing from grass and croplands to forest, there are two competing effects of land cover change on climate: an albedo effect which leads to warming and an evapotranspiration effect which tends to produce cooling. It is not clear which effect would dominate. We have performed simulations of global land cover change using the NCAR CAM3 atmospheric general circulation model coupled to a slab ocean model. We find that global replacement of current vegetation by trees would lead to a global mean warming of 1.3°C, nearly 60% of the warming produced under a doubled CO2 concentration, while replacement by grasslands would result in a cooling of 0.4°C. It has been previously shown that boreal forestation can lead to warming; our simulations indicate that mid-latitude forestation also could lead to warming. These results suggest that more research is necessary before forest carbon storage should be deployed as a mitigation strategy for global warming.