Although protected areas, or “parks”, are among the leading policy tools used to stem tropical deforestation, rigorous evaluations of their effectiveness—that is, evaluations that control for their tendency to be sited in remote areas with relatively little deforestation—have only recently begun to appear. Important open questions concern the link between the stringency of protection and park effectiveness. How do mixed-use parks that allow sustainable extractive activities perform relative to strictly protected parks? And what types of mixed-use management perform best? In addressing these questions, it is particularly important to control for nonrandom siting, since different management regimes tend to be sited in areas with different preexisting characteristics. To date, most rigorous studies of this issue have focused on scores of parks in one or multiple countries, a strategy that in principle could be undermined by unobserved park heterogeneity. This paper uses high-resolution 2001–2006 land cover data derived from satellite images along with statistical techniques that control for nonrandom siting to examine the relative effectiveness of strict and various mixed-use protection strategies in a single large park: the two-million-hectare Maya Biosphere Reserve in Guatemala. Our results comport with the emerging consensus that on the whole, mixed-use protection in this park has been more effective in stemming deforestation than strict protection because of the performance of forest concessions within the multiple-use zone. However, we also find that mixed-use protection has had smaller, more heterogeneous effects than indicated by simple methods that do not control for nonrandom siting.