Reasoning algorithm of single-value fuzzy causality diagram (SFCD) cannot directly apply to multi-value fuzzy causality diagram (MFCD). So it is necessary to study reasoning algorithm of MFCD. Firstly, with the discussing of reasoning problem of MFCD in this paper, a guideline to solve the problem is introduced, and a normalization method of fuzzy probability of event state is proposed to preprocess data. Secondly, a reasoning algorithm of MFCD based on unitizing coefficient is proposed to deal with the reasoning of MFCD. Lastly, an example about fault diagnosis of a steam generator in the nuclear power plant demonstrates the effect of the reasoning algorithm of MFCD, and the result is coincident with the fact. The research shows that the reasoning algorithm of MFCD is so effective to solve the problem of MFCD for fault analysis and reasoning, its reasoning process is rigorous, and the result coincides with the reality. Index Terms--Causality Diagram (CD); Fuzzy; Single-value; Multi-value; Reasoning Algorithm; Unitizing Coefficient.