A new complexity metric to evaluate the unpredictability of the chaotic pseudorandom sequences based on the fuzzy relationship entropy (F-REn) is proposed in the view of maximal randomness of the sequences with arbitrary length. On this condition,two basic properties of F-REn are proved. Simulations and analysis results show that, the proposed F-REn works effectively to discern the changing complexities of the chaotic pseudorandom sequences, and compared with complexity metric based on the approximate entropy(ApEn) and symbolic dynamics approach , F-REn works have obvious advantages in the applicability of symbolic space, the sensitivity of vector dimension and the robustness of resolution parameter.