The essence of oil monitoring and fault diagnosis is a multisource information processing process for recognition tasks.With the lubrication and friction systems becoming more complicated and sophisticated,traditional thresholds and expert experience have difficulty in meeting fault diagnosis requirements,the fault classification methods by pattern recognition theories have become an important development direction of oil fault diagnosis.The development actuality of pattern recognition theories for oil monitoring was briefly described from two aspects of feature extraction theories and fault mode judgment theories.The characteristics of various theories were analyzed,and the development trend and problems of pattern recognition theories for oil monitoring were pointed out.