Most　of　intrusion　detection　systems　nowadays　are　not　really　distributed　systems　which　cannot　detect　the　distributed　or　cooperative　attacks　effectively.　In　this　paper,　an　Agent-Based　Distributed　Cooperative　Model　(ADCM)　is　proposed,　which　implements　cooperative　intrusion　detection　through　efficient,　normative　event　messages　exchange　among　Logic　Detection　Domains　(LDD).　Some　specific　detection　agents　are　also　presented　which　are　independent　separately,　while　they　can　communicate　and　cooperate　with　one　another　to　take　actions.　The　ADCM　improves　the　ability　of　error　tolerance　and　cooperation　without　degradation　of　efficiency.　Prototype　of　a　distributed　intrusion　detection　system　based　on　ADCM　and　the　extended　intrusion　detection　message　exchange　format　is　completed,　which　proves　to　be　powerful　as　expected　in　detecting　intrusions.