In Sahel, land and resources are used, according to the seasons, by various actors with interests sometimes difficult to conciliate. This spatial competition for the access to the natural resources induces a set of processes leading to overgrazing and conflicts for the access to water. In such a complex agro-sylvo-pastoral context, acquiring and formalizing knowledge about the practices and rules which govern the system are essential to help and accompany the rural populations towards the sustainable management of their environment. Among the new modelling tools coming from the field of Artificial Intelligence, Agent-Based Models (ABMs) are now widely used to tackle the issue of integrated natural resource management. The increase in popularity of ABMs is partially due to their intelligibility, not only by scientists from various research fields, but also by stakeholders. It then becomes easier to build an understandable representation of the system as an "artificial world" and to perform simulations in order to collectively test and discuss various scenarios about the resource management. This paper describes an ABM which has been designed to formalize the interactions between the biophysics dynamics of the natural resources and the socio-economic factors driving the land-use dynamics around the drilling of Thieul village in the sylvo-pastoral area of Ferlo (Senegal). The first results show that rainfall still plays an extremely significant role in the pastoral system despite the drilling of boreholes. Paradoxically, steady rainfall over long periods of time can lead to negative effects on the relations between herders and farmers. (c) 2005 Elsevier Ltd. All rights reserved.