To solve the problem of radiation-source target localization for a single UAV(unmanned aerial vehicle) in an urban environment, a new trajectory optimization algorithm for bearings-only target localization based on the environment prediction method was proposed. Interacting multiple model methods coupled with the extended Kalman filter was used to estimate the target localization in the line-of-sight and non-line-of-sight mixed environment. Based on the estimated target location and urban geographic information system, the electromagnetic signal occlusion region and the multipath interference region were calculated by using the line of sight tracking method. Under the framework of receding horizon method, the UAV prediction trajectory was generated, so as to maximize the Fisher information matrix determinant as the orientation positioning evaluation criterion. Considering the influence of building obstacles and their occlusion and reflection effects in the localization process, the UAV was controlled to choose the optimal heading flight.The numerical simulation results show that the trajectory optimization algorithm enables the UAV to perform high-precision bearings-only target localization in the complex environment containing obstacles, signal occlusion, and multipath interference. The algorithm provides a new way to solve the problem of bearings-only target localization for single UAV in an urban environment.