There is widespread recognition that forestry carbon credits can reduce the net emissions of carbon into the atmosphere. Designing systems to sequester carbon, however, has proven difficult due to a number of efficiency issues, including leakage. Leakage occurs when policy makers develop carbon projects in specific places which protect some parcels of land, but leave other parcels of land unprotected. This analysis uses a newly developed model of global land use change from an established forestry and land use model, described in Sohngen et al. (1999); Sohngen and Mendelsohn (2003); and Kindermann et al. (2008). To assess leakage we estimate carbon under storage under one scenario where the world is awarded carbon credits and another where tropical developing nations are awarded the credits. We focus our results on several regions, namely Brazil, the rest of South America, Sub-Saharan Africa and Southeast Asia. Carbon prices are assumed to be constant, and range from US$0 tC to US$900 tC. The model adjusts global land uses to these specific policies, and leakage is assessed by comparing carbon gains within the project areas to net global changes in carbon. A number of policy relevant results emerge. First, the estimates indicate that leakage ranges from 2% to more than 14%. Second, as carbon credits increase, leakage decreases across the world.