For non-real time multimedia systems, we present a fuzzy approach to replacing the faulty module. After analyzing the nature of the random and pseudo-random test sequences applied to a module under test, we obtain the aliasing fault coverage between the random and pseudo-random sequences. The activity probability features of intermittent faults in the module under test are discussed based on the Markov chain model. Results on real examples are presented to demonstrate the effectiveness of the proposed fuzzy replacement approach