With the potential expansion of forest conservation programs spurred by climate-change agreements, there is a need to measure the extent to which such programs achieve their intended results. Conventional methods for evaluating conservation impact tend to be biased because they do not compare like areas nor do they account for spatial relations. We assess the effect of a conservation initiative that combined designation of protected areas with payments for environmental services to conserve overwintering habitat for the monarch butterfly (Danaus plexippus) in Mexico us a spatial-matching estimator which matches covariates among polygons and their neighbors, to. We measured avoided forest loss (avoided disturbance and deforestation) by comparing forest cover on protected and unprotected lands that were similar in terms of accessibility, governance, and forest type . Whereas conventional estimates of avoided forest loss suggest that conservation initiatives did not protect forest cover, we found evidence that the conservation measures are preserving forest cover. We found that the conservation measures protected between 200 and 710 ha (3-16%) of forest that is high-quality habitat for monarch butterflies, but had a smaller effect on total forest cover, preserving between 0 and 200 ha (0-2.5%) of forest with canopy cover >70%. We suggest that future estimates of avoided forest loss be analyzed spatially to account for how forest loss moves across the landscape. Given the forthcoming demand from donors and carbon financiers for estimates of avoided forest loss, we anticipate our methods and results will contribute to future studies that estimate conservation impact .