Dozens of countries have decentralized at least part of their natural resource policies over the last two decades. Despite the length of time that these policy experiments have been in force, there is little agreement about their effectiveness. We argue that part of this ambivalence stems from three limitations of extant studies, suggesting that future studies of decentralized natural resource governance should consider a combination of: (1) variation in the local institutional context; (2) the fit between the reform and other public policies; and (3) more adequate outcome measures for decentralized resource governance. After developing such an approach, we posit that varying forest conditions depend on the moderating effects that local institutions have on the socioeconomic and biophysical drivers of environmental change. Analyzing data from interviews and remotely sensed images from 30 municipalities in the Bolivian lowlands, we find that the local institutional performance affects unauthorized deforestation directly and indirectly, but detect no effects on either permitted or total deforestation.