Construction projects are subject to numerous risks that could have consequences on project achievement. They involve numerous stakeholderswhose interests and demands need to be considered in the managerial decisionmakingprocess, to ensure the success of the project. Each stakeholder has his own risks and perspective on project risks. Furthermore, risk management is a dynamic process throughout the project life cycle. On one hand, risks change along the different phases of the project, on the other hand, multiple interactions have been identified between risks. For these reasons, it could be difficult to model project risks in order to manage them. In this paper, a multi-agent model (SMACC) is proposed to study this complex system. By using multi-agent simulations coupled with a stochastic approach, risk impacts can be evaluated for each stakeholder and for the whole project. These simulations make possible to test different risk mitigation strategies in order to measure their interest and then to support risk management decisions.