Global efforts to reduce deforestation rely heavily on protected areas and land use restrictions. The effect of these restrictions on local communities is currently the subject of heated debate among conservation and development experts. Measuring the social impacts of protected areas is difficult because the effects cannot be isolated from other factors, given the nonrandom placement of protection. We address this problem by applying a quasi-experimental approach to establish the counterfactual (“what would have been the socioeconomic outcome if a protected area had not been established?”). We use matching methods to measure the impacts of pre-1980 protected areas in Costa Rica on socioeconomic outcomes in 2000. In 2000, neighboring communities near protected areas were substantially poorer than average. However, after controlling for pre-protection characteristics associated with both protection and economic growth, the results indicate that poverty declined as a result of protection. Although the statistical significance of this decline is moderately sensitive to potential hidden bias, the results emphatically do not support a hypothesis that ecosystem protection, on average, exacerbates poverty. In contrast, conventional empirical methods implied erroneously that protection had negative social impacts, suggesting that failure to control for confounding factors or baselines can lead to substantially inaccurate estimates.