An optimization method was put forward to adaptively generate diagnostic strategies of complex feedback loop systems. Firstly, the exsiting cycle searching algorithms were improved to beamore effici ent algorithm that could find all feedback loops faster. Secondly, three breaking strategies at different situations were presented. After that,anal gorithm to generate efficient diagnostic strategies was presented. This algorith memployed the information gain as the heuristic function of test sequencing and achieved a nearly optimal diagnositc strategy by interative computation. Finally, the main steps of the method were provided, and the method was validated byan example. The experimental results show that the presented method efficientlys olves the diagnostic problem of feedback loop systems.