Abstract: In order to solving the low precision problem of exsiting nonlinear filtering algorithms when used for moving multi-platform passive tracking, a truncated unscented Kalman filtering(ITUKF) is poposed. The proposed algorithm truncates the probability density function (PDF) of the measurement noise and the prior PDF of the state to make them have a bounded support. Combing with the original prior PDF of the state, a mixture prior PDF of the state is designed. The unscented transformation (UT) is applied to each of the prior PDF to calculate first two moments of the corresponding posterior PDF and then these moments are merged to form the final state estimation. Simulation results indicate that compared with several typical nonlinear filtering algorithms, the TUKF algorithm can effectively improves the tracking performance.