Aiming at poor anti-error performance in the existing algorithms for RS codes, a blind recognition algorithm based on non-zero-mean-ratio is proposed. The algorithm constructs the analysis matrix by transforming the intercepted RS code sequence into GF(2 ^m ) symbols. The finite-field Gauss-Jordan elimination algorithm is used to obtain the non-zero-mean-ratio of the analysis matrix and identify the code length, symbol number and the primitive polynomial. Finally, the Galois field Fourier transform is used to complete the information bit length and generator polynomial recognition. Simulation results show that the proposed algorithm can effectively identify all the coding parameters of the primitive RS code and shorten RS code. The anti-error performance is better, and the relationship among the recognition performance and the information bit length and the code length is given. As the code length and information bit length increase, the recognition performance gradually decreases.