The ghost problem exists in multi-DF station multi-target crossing-location. The number of ghost increases rapidly with the number of DF station and target. It has a very low correct association probability when the DF error is comparatively large. Aimed at this problem, the improved algorithm combined with TDOA information to eliminate ghost was presented. On the basis of crossing location using azimuth angle information, the data comparability between DF stations was computed by the algorithm combined with TDOA information, and associated data by the rule of largest data comparability. The results of the simulation experiment show that this algorithm has higher correct association probability and can eliminate location ghost more effectively.