Agent-based models (AMB) allow to conceive social systems as the result of individually-acting agents. When they are applied to agriculture, they can simulate the fundamental behaviour at the micro-level of the individual farmers, without the need of aggregating them in “representative” agents. RegMAS (Regional Multi Agent Simulator) is an open-source spatially explicit multi-agent model framework specifically designed for long-term simulations of effects of policies on agricultural systems. Using iterated conventional optimisation problems as agents' behavioural rules, it allows for a bidirectional integration between geophysical and social models where spatially distributed characteristics are taken into account in the linear programming problem of the optimising agents as individual resources. The model is applied to asses the impact of the Health Check, the imminent further Common Agricultural Policy (CAP) reform, on farms structures, incomes and land use in a hilly area of a central Italian region (Marche). Our results suggest that the Health Check, while increasing the farmer profit net of CAP support, is substantially neutral on the overall farmer incomes, also through a reduction of the off-farm labour. Neverless, a limited negative effects may arise in the farms numerousness, with the consequence of a land abandonment that is noticeable only on mountain areas, where distances between farmers are greater.