Agroforestry systems are complex assemblages of ecosystem components, each of which responds to climate. Whereas climate change impacts on crops grown in monocultures can reasonably well be projected with process-based crop models, robust models for complex agroforestry systems are not available. Yet, impact projections are needed because of the long planning horizons required for adequate management of tree-based ecosystems. This article explores available options for projecting climate change impacts on agroforestry systems, including the development of process-based models, species distribution modeling, climate analogue analysis and field testing in climate analogue locations. Challenges and opportunities of each approach are discussed.