With the development of the aviation industry, methods for aero-engine fault diagnosis have become increasingly intelligent and accurate. In this paper, we proposed a method that combines fuzzy clustering, rough sets and support vector machine (SVM). First, a fuzzy C-average clustering algorithm was applied to discretize the continuous data. Then, we used the knowledge discovery theory of rough set to reduce the decision table under the premise of keeping the table‘s attribute and dependencies between conditions attributes unchanged. We used the SVM to study samples to obtain the optimal hyperげplane decision function. Finally, we used the diagnosis faults based on these characteristics for the data processing of small samples. The instance validation results of the aero-engine performance parameters showed that our method had improved ability to diagnosis aero-engine faults and could greatly shorten operation time without affecting the diagnostic rate. Thus, the proposed algorithm is both practical and accurate.