Human error is one of the most critical factors leading to aviation unsafe incidents/ accidents. A human reliability assessment model for aircraft operating based on an FBCREAM method is proposed to predict human error probability in cockpit. It can be further used to reduce human errors and increase aviation safety. Considering the aviation scenarios and human factors in cockpit, the performance shaping factors in the original CREAM (Cognitive Reliability and Error Analysis Method) are adjusted to more reasonably characterize pilot's working situation in flight. Then, the fuzzy logic method is employed to model the uncertainty and ambiguity of performance shaping factors. The Bayesian network model is constructed with the membership of the performance shaping factors as the input parameter and the membership of the human error mode as the output. Finally, human error probability can be calculated by the defuzzification method. An experiment of the flight mission under the single-fault fault scenarios is performed. The results show that the proposed model can accurately calculate human error probability in unsafety incident. It can provide an effective tool and important support for aviation safety assessment.