Abstract: In the system reliability analysis, in addition to the common random uncertain information, also there is fuzzy uncertain information. On the basis of traditional reliability and fuzzy theory, a kind of reliability calculation method which could handle random uncertain information and fuzzy uncertain information comprehensively is proposed, and reliability calculation model is got, which could process the probability density function and membership function simultaneously. What’s more basic on the improved evidence theory, two kinds of information integration methods are proposed by integrating reliability information from different sources. The study results show that：in the case of both randomness and fuzziness existing, this reliability model could handle random information and fuzzy information more appropriately, besides a higher accuracy, it could also reflect reliability calculating mechanism dynamically; information integration method could process reliability information from different sources in a better way, therefore it is a useful method of calculating reliability form multi-source information.

Key words: Evidence theory, Fuzzy reliability, Information integration, Membership function, Uncertain information