The invention provides a bank client abnormal behavior analysis method. The method comprises the steps that by combining a modern information technology and a professional field development technology, all client behavior data are integrated and processed, the range of behavior influence is studied, scientific modeling is conducted on client behavior information on the basis of an SVM algorithm and a rough set, and clients with abnormal behaviors can be reminded in time. Through the mode, client behavior data can be dynamically and scientifically managed, so that the client service level is remarkably improved. The quick response and adaptability of the banking industry are enhanced. The operation cost and risk of the banking industry are reduced, and great indirect economic benefits and social benefits are brought.