In this article, we propose an original method for providing fault tolerance in multi-agent systems. Our method focuses on building an automatic and adaptive replication policy to solve the resource allocation problem of determining where agents must be replicated to minimize the impact of failures. This policy is determined by taking into account the criticality of the agents and the reliability of the machines. We propose then different heuristics for the allocation of the available resources. Some measurements assessing the effectiveness of our approach are also presented.