A novel method is presented to recognize warheads in the ballistic missile defense. The normalized correlation coefficients and further the matching scores between range profiles (RPs) of warhead are first calculated. By introducing a threshold, the warhead and decoy can be separately classified. The distribution of these results is then modeled as a mathematical model. Based on such distribution, both the recognition rate of warhead and the false alarm rate of decoy can be obtained. The experimental results using the simulated data demonstrate the effectiveness of our method.