A multiple hypothesis tracking (MHT) algorithm based on multi-feature fusion is presented in this paper to counter range deception jammings. Sparse decomposition coefficients and bispectrum features are extracted to distinguish the targets and the jammings. A two-stage fusion structure using neural network and Dempster-Shafer evidence theory is designed to implement multi-feature fusion so as to get the classification probabilities of the measurement being target originated or jammer originated. Then, a multiple hypothesis tracker accounting for the deception jamming is presented. The hypothesis probabilities are derived to incorporate the classification probabilities so that the probability of correct data association will be increased. Simulation results show that the proposed approach has better robustness than the amplitude aided MHT while with comparable tracking performance in terms of track continuity and track accuracy.