Since the end of the 70's, various forest policies have been implemented in the Sahel to address the problem of ecosystem degradation and the increasing need for fuelwood energy. In Niger, a new environment-oriented policy is now being implemented, in conjunction with the decentralization process and the transfer of specific forest user-rights to the local population. Incentives, regulations and new institutional arrangements are designed to take account of the forest ecosystem and forest uses. How can the robustness of policy rules be assessed when these rules are implemented in a dynamic ecological and economic environment? According to what criteria can the effects of these rules be analysed? How can a reorganization of the economic sector enhance the effectiveness of the rules? We choose to explore these issues using a multi-agent modelling approach, based on the interactions between the forest resource, the economic agents (from harvesters to wood traders and urban fuelwood consumers) and the policy rules. We simulate scenarios for the various incentive measures which are implemented. The agent-based model offers an ecosystem-based approach to incentive measures and allows us to demonstrate how institutional arrangements, rules and rule compliance are affected by the behaviours of local actors and their interactions. The model relies on a case study of institutional change in Niger, where the new policy has led to a reorganization of ecosystem exploitation. On the basis of an initial model presented here, we illustrate how the approach based on multi-agent models can be used to study the implementation of ecosystem management rules.