An ultra-short baseline positioning algorithm based on region-division adaptive particle swarm optimization is presented in order to reduce the underwater acoustic positioning error caused by inaccurate sound speed. This algorithm regards sound speed as variables and utilizes redundant information to build the positioning model. For the shortcoming that the standard particle swarm algorithm is slow in convergence speed and easy to fall into local optimum, the region-division method is utilized to dynamically adjust the inertia weight and learning factors of each particle to reach the balance between optimization ability and convergence speed, and the adaptive mutation operation is introduced to avoid the population falling into local optimum. Finally, simulation results show that the proposed algorithm can effectively improve the positioning accuracy of the ultra-short baseline system with unknown sound speed.