In bearings-only target localization, observer path have an important effect on the accuracy of target localization. Based on the analysis of Fisher information matrix, an optimal approach to designing observer path is proposed. Observer path derived from the step-by-step optimal algorithm only depends on observer's initial states and state constraints. The optimal path resulted with step-by-step algorithm is demonstrated by computer simulation. Compared with traditional methods, our algorithm is not limited to the number of observations as well as terminating conditions.