We summarize existing theoretical claims linking poverty to rates of deforestation and then examine this linkage empirically for Costa Rica during the 20th century using an econometric approach that addresses the irreversibilities in deforestation. Our data facilitate an empirical analysis of the implications for deforestation of where the poor live. Without controlling for this, impacts of poverty per se are confounded by richer areas being different from the areas inhabited by the poor, who we expect to find on more marginal lands, for instance less profitable lands. Controlling for locations’ characteristics, we find that poorer areas are cleared more rapidly. This result suggests that poverty reduction aids forest conservation.