4 October 2018Since the beginning of the REDD+ mechanism, hundreds of projects have emerged around the globe. Much attention has been given to REDD+ projects in the literature, but the con- ditions under which they are likely to be effective are still not well known. In particular, the location bias concept states that projects are more likely to be implemented in remote areas, where development pressure is low, and hence questions their additionality. In this article, we revisit this concept, trying to assess the process of REDD+ projects implementation and its influence on project additionality. First, a simple theoretical model shows that project imple- mentation is influenced by the type of project proponent, which appears to be a good proxy for its objectives, whether oriented toward environmental impacts, development impacts, or external funding. Our results suggest that (1) the project proponents objective and local institutions may lead the project proponent to select a community with low development potential, and (2) the selection of a low-development potential, which is frequently presented as a location bias, does not necessarily preclude additionality. Those predictions are empirically tested on a sample of six REDD+ projects in Brazil. We propose an empirical analysis of the location choices and estimate additionality in the first years of implementation using impact evaluation techniques. The results confirm the predictions of the model.