The key problem for reentry vehicle (RV) discrimination during the reentry phase is to estimate the mass-to-drag ratio of RV precisely and rapidly. Aiming at the nonlinear characteristic of reentry process, three filters：as spline Kalman filter, extended Kalman filter and a new extended Kalman filter based on “unscented transform” are studied. The comparison of estimating performance for these filters are provided through simulating experiment with terms of estimating precision and converging speed. Experimental results show that the new EKF based on “unscented transform” has the best performance on the estimation of mass-to-drag ratio.