Oil monitoring technology is a useful method in condition monitoring and fault diagnosis for the machine, especially for low-speed, heavy-load, reciprocated and lubricated diesel engine equipment. But it is difficult to implement intelligent diagnosis because monitored information lacks logical relationship in oil monitoring. To solve this problem, the theory and method of case-based reasoning is adopted for the data processing and fault analysis in oil monitoring with an multi-agent system structure. Detailed definitions of agents in the system were proposed, and the multi-agent system framework was established finally. Multi-agent mechanism brings flexible for case based reasoning. It enhances the capability of solving complicated question in new system, and overcomes the shortcoming of the fault knowledge difficult to update in traditional systems.