trialspecies,therelativemeritsofdifferentapproachesto ensure the long-term persistence of those species remain highly contentious. Most would agree, however, that both establishing protected areas and exercising some form of restraint on extraction of forest resources are among the most effective of all viable conservation measures. Deforestation, wildfires, logging, and hunting are among the leading drivers of species losses in tropical forests, and de facto or de jure protection from these threats can be conferred by either effective enforcement of regulations or physical remoteness. Attempts to assess conservation success of protected areasatlargescaleshaverestedprimarilyonconventional use of remote sensing to quantify spatial or temporal differences in rates of change in land cover due to deforestation and wildfires, rather than on empirical demographic or community-level metrics (Gaston et al. 2008). Q2 However, the former approaches fail to detect most types of subcanopy anthropogenic disturbances that also result, directly or indirectly, in species losses (Peres et al. 2006). Moreover, the effects of habitat loss and degradation on population extirpations and declines are nonlinear, so vegetation cover alone is rarely a robust proxy for the viability of terrestrial biotas. Remotesensing data show vast tracts of apparently intact tropical forests, but in reality levels of hunting and other forms of extraction in these areas are often unsustainable (Peres & Lake 2003). Fundamental questions yet to be answered include whether ostensibly intact protected areas retain full complements of forest species and how the extent of cryptic patterns of disturbance is related to human population density in both protected and unprotected areas. I considered the global to regional emergence of sustainable-use reserves, emphasizing the world’s largest tropical forest region, Amazonia. Sustainable-use reserves often have intermediate levels of disturbance, so I examined the degree of use of natural resources by resident communities and used human population density as a proxy for level of extraction. In both protected and unprotected areas, I also estimated responses of game vertebrate assemblages to hunting on the basis of the relative biomassextractedfromasubsetoftheforestfauna.Iused analysesofcovariance(ANCOVA)toexaminetheassociationbetweenhumandensityandgamebiomassharvested across different reserve categories. Finally, I considered the long-term capacity of sustainable-use forest reserves to maintain populations of all resident species.