In this paper, a novel and cooperative approach is exploited introducing a self-organizing engine to achieve high reliability and availability in multiagent systems. The Adaptive Multiagent Systems theory is applied to design adaptive groups of agents in order to build reliable multiagent systems. According to this theory, adaptiveness is achieved via the cooperative behaviors of agents and their ability to change the communication links autonomously. In this approach, there is not a centralized control mechanism in the multiagent system and there is no need of global knowledge of the system to achieve reliability. This approach was implemented to demonstrate its performance gain in a set of experiments performed under different operating conditions. The experimental results illustrate the effectiveness of this approach.