We have constructed a comprehensive database of 117 spatially explicit econometric studies of deforestation published in peer-reviewed academic journals from 1996-2013. We present a metaanalysis of what drives deforestation and what stops it, based on the signs and significance of 5909 coefficients in 554 multivariate analyses. We find that forests are more likely to be cleared where economic returns to agriculture and pasture are higher, either due to more favorable climatological and topographic conditions, or due to lower costs of clearing forest and transporting products to market. Timber activity, land tenure security, and community demographics do not show a consistent association with either higher or lower deforestation. Population is consistently associated with greater deforestation, and poverty is consistently associated with lower deforestation, but in both cases endogeneity makes a causal link difficult to infer. Promising approaches for stopping deforestation include reducing the intrusion of road networks into remote forested areas; targeting protected areas to regions where forests face higher threat; tying rural income support to the maintenance of forest resources through payments for ecosystem services; and insulating the forest frontier from the price effects of demand for agricultural commodities.JEL Codes: Q15,Q23,Q24