The study of rodent animal’s social behavior has a key role in the field of biology and medicine, and it’s a challenge to track multiple animals individually in the group experiment. The trajectories swap is a common problem in multiple mice tracking, and it is difficult to be solved by single vision method. This paper discusses a hybrid tracking method using RFID (Radio Frequency Identification) and computer vision. It first adopt the nearest neighbor method to link the positions calculated by recorded images of every target and form the original trajectories. Then location data recorded by the RFID system are fused with the trajectories identified by the vision system. Once the trajectories swap appears, the method uses the RFID location data to correct the identities of the trajectories. This paper presents the design of the vision tracking and RFID location system, and the data fusion algorithm of vision and RFID. It also presents the performance of the method with experimental tests.