Abstract: At present, there is still a lack of effective identification methods for oil and gas pipeline surface defects. To solve this problem, based on metal magnetic memory technology, a multi-characteristic statistical identification method for the identification of pipeline corrosion defects and early stress concentration defects was established through the magnetic memory test data obtained from pipeline test pieces with circular-hole corrosion defects and elbow bend stress concentration defects. Under laboratory conditions, the defect types of the three test pieces were identified by this method. Under the site conditions of the oil field, non-blind test and blind test were respectively conducted to verify the defect types of two pipelines. The identification accuracy rate is above 80%, which indicates that the established defect type identification method is effective and has high identification rate for the identification of pipeline corrosion defects and early stress concentration defects, so it can provide certain reference for the research in this field.