The interactions between natural forest dynamics and social dynamics have to be taken into account when managing the use of forest resources. We have developed simulation models to improve our understanding of this complex system of interactions. Models of multiagent systems are effective tools for studying the dynamics of complex adaptive systems. We have developed several simulation models in order to study the use of forest resources. This paper presents a model designed to understand the interaction between fuelwood consumption and landscape dynamics. The hypothesis put forward suggests that fuelwood consumption can explain the landscape changes that occur in the Kayanza region of Burundi. The second hypothesis is that a sustainable use of resource must keep steady the fuelwood consumption level per capita over time. A preliminary map was outlined: agents use fuelwood, have access to different parts of space and have the capacity to exchange use rights. The population increases and agent migration from overpopulated areas to unoccupied plots - is simulated. The impact of changing rules on foraging, exchange and access is then observed on a landscape level. The model describes here the behaviour of different agents (farmers, local consumers, exporters and traders). The impact of their behaviour and interactions are evaluated on different scales, ranging from the individual plot to the forest and the landscape. Models and multiagent systems can effectively represent processes that occur at levels of varying complexity and simulate their interactions so that landscape dynamics can be understood from the bottom up.