Abstract: In order to improve the accuracy of the target recognition under the condition of conflicting evidence, we propose a target recognition algorithm improving the order of the evidence fusion by using the comprehensive reliability of the evidence. The new algorithm determines the parameters of the conflict measurement based on the differences between every two pieces of evidence, and gives the formula for calculating the comprehensive reliability of evidence. On this basis, the order of evidence fusion is optimized, and the contradictory information contained in conflicting factors is redistributed by weight among relevant focal elements. The numerical examples and the simulation results show that the new algorithm can effectively deal with different types of conflicting evidence. Especially in the case of highly conflicting evidence, compared with the direct fusion algorithm of multi-dimensional evidence, the correct target recognition probability of the new algorithm is improved by 0.15~0.24 points, and the time spent is less than one-tenth of that of the direct fusion algorithm of multi-dimensional evidence. Therefore, the proposed algorithm is a target recognition algorithm with high stability and less time spent.