Abstract: Aiming at the problems of insufficient utilization of single-physics-field information, strong non-uniqueness of anomaly interpretation, and difficulty in lithology identification under complex geological conditions in traditional gravity and magnetic anomaly interpretation, this paper proposes a comprehensive interpretation method based on multi-attribute fusion and adaptive classification. This method comprehensively utilizes gravity anomalies, magnetic anomalies, and their variation characteristics. It further introduces the horizontal gradient magnitude for boundary feature recognition, performs unified processing and fusion expression of multi-source data, and achieves information integration and automatic classification recognition within the same feature space. Through model tests under different physical property contrasts, burial depth variations, and anomaly superposition conditions, the experimental results show that this method can stably distinguish different types of geological bodies and exhibits good boundary recognition capability and anti-interference ability. The method is applied to the processing of gravity and magnetic data in the Ordos Basin. The results indicate that this method can effectively achieve anomaly zoning and lithology identification. Different clustering categories show significant differences in gravity and magnetic anomaly characteristics, which correspond well to the regional basement lithology distribution, providing a new technical approach for the comprehensive interpretation of gravity and magnetic data in complex areas.