Abstract: To address the challenges posed by complex and frequently changing motion patterns of underwater targets in practical applications, this paper proposes an underwater target localization and tracking method based on the interactive multiple model (IMM) filtering framework. The method integrates two typical motion models of underwater targets—constant velocity (CV) and constant turn (CT)—and fuses observations from different time instants to achieve dynamic model switching and joint state estimation via the IMM framework. Furthermore, estimation under two types of acoustic measurement conditions, namely direction of arrival-time of arrival (DOA-TOA)、time difference of arrival (TDOA) and DOA-TOA-TDOA, an IMM-unscented Kalman filter (UKF)-based localization and tracking algorithm is designed using the combined motion model. Simulation results demonstrate that the proposed method outperforms traditional single-model algorithms in terms of adaptability to multiple motion patterns and localization accuracy. In addition, the effectiveness and feasibility of the algorithm are further validated through processing and analysis of real underwater acoustic observation data in practical engineering scenarios.