The economic modernization of the Amazon fostered by the Brazilian military government during the 1960s and 1970s was largely realized without taking into consideration the presence of local households which lived from the extraction of forest products (mainly non-timber). When they began to be expulsed, a political resistance, often guided by the Catholic Church, appeared as well as the creation of unions based on traditional identities, especially that of rubber tappers. During the 1980s, these unions made a strategic alliance with the ecologist movement which started to consider traditional populations, whose lifestyle depended on the forest, as allies for the protection of the Amazon rainforest. The movement gained a decisive momentum at the end of the decade by putting forward new proposals of land tenure for traditional populations, grounded on collective land rights. This strategy has been very efficient during the 1990s and 2000s, during which about 1,300,000 km 2 of rainforest were set apart and reserved for the use of " traditional communities " under a variety of legal status. But it has also led to mix under the same " collective " etiquette and principles a number of different ways of using and managing land and natural resources. This assumption however should be nuanced by a careful analysis of the resource management systems existing in each case, for they are in general complex and mix varying proportions of individual and collective decisions. The aim of this paper is to explore this question using the example of the Chico Mendes agro-extractive settlement (PAE-CM), inhabited by about 100 rubber tapper families and symbolic of the political struggle of traditional populations in the Amazon for being the home of the rubber tapper leader Chico Mendes assassinated in 2 Francois-Michel Le Tourneau and Bastien Beaufort 1988. Applying Ostrom " design principles " , we try to catch what are the local institutional arrangements and to see if they suggest collective or individual management , and what the boundaries between both categories are. As a conclusion, we find that the PAE-CM's system is much less collective than expected, and also very much controlled by external authorities, in a logic pretty much away from the idea of a CPR system. This finding is useful to understand the shortcomings in the actual management of the PAE but also to foresee difficulties which will probably arise in the management of many of the areas which have gained collective land rights or collective management statutes in the Amazon.