Introduction: The porcine respiratory and reproductive syndrome (PRRS) is a major viral disease in swine production. In most cases, farmers decide whether to control this disease or not within their herd on a voluntary basis. Nevertheless, individual decisions have an impact on the risk for other farmers to be infected. Since some farmers are grouped in associations/geographical areas, it is relevant to investigate how a group of farmers can coordinate individual decisions, implementing incentives for disease management. The objective of this study is to develop a framework to propose an optimal strategy to be used by the decision maker of an association, strategy which consists on rules to define at each time step which incentives should be used. Materials and Methods: We describe an approach to propose control strategies which are adaptive to the evolution of the epidemiological situation over time. We assumed that the collective decision-maker should at each time-step select the incentive to optimise a criterion, for example the minimisation of the total cost at the group level (incentive, control and disease costs). The decision-maker can choose among various incentives levels, ranging from cheap no-incentive to costly incentives. We assumed that the decision-maker knows how the farmers would react to its incentives. A Markov decision model was defined including stochastic compartmental models representing PRRS virus spread within a group of farms among which some are PRRS virus positive. For each incentive level, the proportions of farmers implementing controls measures was introduced into the model avoiding the formalisation of individual decisions. The model was solved to produce a strategy optimising the total cost at the group level. For comparison, we simulated the spread of the PRRS virus within the group of herds in 2 situations : (1) when the collective decision-maker uses the optimal strategy and (2) when no incentive was used by the collective decision maker over time . Results: We obtained a strategy corresponding to a guideline indicating at each time step the action to use according to the observed epidemiological situation. This guideline is translated into a decision-tree. When using the optimal strategy, various levels of incentive are used over time inducing an average total cost at the group level lower than if we systematically used each incentive level. Conclusion: We propose a complex strategy optimising the total cost which is translated in simple rules (“if-then”). While optimising the total cost, the model can be extended to consider also an objective in terms of prevalence decrease