The traditional event tree analysis uses a single probability to represent each top event. However, it is unrealistic to evaluate the occurrence of each event by using a crisp value without considering the inherent uncertainty and imprecision a state has. The fuzzy set theory is universally applied to deal with this kind of phenomena. The main purpose of this study is to construct an easy method to evaluate human errors and integrate them into event tree analysis by using fuzzy concepts. A systematic fuzzy event tree analysis algorithm is developed to evaluate the risk of a largescale system. A practical example in of offshore oil pipeline is used to demonstrate this procedure.