The feeder automation algorithm has the typical characteristics of discrete-type software. The reliability question has all along been one of the key restraining factors for large-scale application， being affected by various kinds of interference factors， including communication failures between system and site terminal unit， switch equipment refusing action， feeders changing its operation mode， and so on. This paper presents a failure-free evaluation test method for feeder fault location， isolation and restoration（FLISR） algorithm of reliability， and combines with Bayes identification test frequency selection method to determine test sample space. By averaging the test sample results on the user side of the contribution degree， the algorithmic reliability of quantitative score is obtained. Finally， an actual site of intelligence distributed feeder automation system is taken as an example for test， which verifies the effectiveness of the proposed reliability evaluation method.