Ethnic minorities living in the highlands of northern Thailand have long been accused of degrading the upper watersheds of the country's major basins. In the nineties, the Thai government reinforced his environmental policies and further restricted their access to farm and forest resources. In the meanwhile, the policy framework also favoured decentralization and public participation. This contradiction resulted in an increasing number of con icts over land-use between local communities and state agencies. This situation underlines the need for adapted participatory methodologies to facilitate the coordination of multiple stakeholders with competing interests. Companion Modelling (ComMod) is one of them. When drawing the lessons from many past participatory projects, several authors highlight their limited impact due to the lack of support at higher institutional levels. Moreover, because of a lack of attention to the local socio-political situations, the less powerful stakeholders were often left behind. This article discusses the usefulness of an area diagnostic study prior to the launch of a ComMod process to avoid such pitfalls and to facilitate genuine communication among stakeholders within and across institutional levels. The article is illustrated by a ComMod experiment conducted in Nan province and is focusing on a con ict between two Yao communities and a recently established national park. We argue that a relatively short but well-structured initial agrarian and institutional analysis to assess the various stakeholders' characteristics, perceptions of the issue to be solved, and interactions is useful to identify the constraints to an equitable outcome of a subsequent participatory process. Such an area diagnosis can also be used to tailor the ComMod process in order to mitigate these constraints. Moreover, an understanding of the initial resource management situation is necessary to be able to monitor changes and to assess the effects of the participatory process.