While the fuzzy power flow is based on fuzzy set theory and has no standard solution till now， the credibility theory， a mathematical branch in the study of fuzzy phenomenon， provides the axiomatic foundation for fuzziness. A general fuzzy power flow algorithm is developed based on credibility theory. The algorithm can not only give the possibility distribution of the power flow results， their expected values and variances， but also calculate the expected values of overloaded equipment's graveness and the system risk. On the one hand， the calculation results of fuzzy power flow can be analyzed and compared directly by the algorithm with improved practicability. On the other hand， it has solid mathematical theory foundation and can be adopted as a benchmark testing method for the power flow. Numerical tests on a 6-node system and an actual regional power grid verify the feasibility and effectiveness of this algorithm.