Byline: Patrick d|Aquino, Christophe Le Page, Francois Bousquet, Alassane Bah The increasing diversity of stakeholders, who often hold differing and conflicting land-use perceptions and strategies, underlines the growing need for innovative methods and tools to support a negotiation process aimed at an enhanced and more decentralised land-use management policy. Land-use management is a complex issue and thus presents an irreducible uncertainty that entails a variety of legitimate perspectives. Due to this complexity, the decision-making process should be incremental, iterative and continuous. This means that the acts of decision making will always be imperfect but will move progressively closer towards a shared satisfactory resolution. However, which tools will best fit such a theoretical starting point? Our team is seeking to develop an innovative form of participatory modelling fitting these hypotheses. As regards regional planning, a novel way to put these assumptions into practice has been under way since 1997 in the Senegal River Valley. This experiment, called 'SelfCormas', is being conducted to test a directed self-designing of modelling tools, Geographic Information Systems (GIS) and Multi-Agent Systems (MAS) by stakeholders and principals. From the initial stages, as little prior design work as possible is executed by the modellers. The process has been organised in the form of participatory workshops, including role-playing games, to support a local decentralisation policy at an operational level (2,500 km2). This sort of accompaniment leads to discussions, appraisals and decisions concerning the planning of land-use management. Just a few months after the first workshops, this process has already culminated in autonomous applications, from local to regional levels.