Network security is one of the weakest and most sensitive areas within an organization, because it requires different strategies to protect vulnerable points as well as the coordination and distribution of efforts in order to encompass all the possible forms and points of computer attacks. In this article a distributed multiagent system is proposed as a tool to protect security in networks with different operating systems and which consequently are susceptible to diverse attacks. The system proposed is composed of heterogeneous autonomous agents with different characteristics in terms of their strategies of intruder detection and protection of primary network systems. It is planned that these agents will base their decisions on the application of different artificial intelligence techniques, such as expert system heuristics, evolutionary algorithms, and neural networks.