The ghost problem exists in multi-DF station multi-target passive cross-location under complicated environment. The number of ghost is increased rapidly and the correct association is more difficult along with the increasing DF station and target. Aimed at this problem, two algorithms of eliminating ghost based on redundant information are presented. By associating data with azimuth angle or TDOA redundant information selectively, high computation burden can be avoided effectively when associating directly with azimuth angle data of all DF stations, based on ensuring certain association probability. Computer simulations show that these two algorithms are effective. Because of using additional TDOA redundant information, the correct association probability of eliminating ghost algorithm based on TDOA redundant information is higher than that of eliminating ghost algorithm based on azimuth angle redundant information when the TDOA error is lower than 0.5ms.The eliminating ghost algorithm based on TDOA redundant information is more adapted to the occasion with high TDOA precision、low space between targets, and big DF error.