A new hierarchical human error classification method is proposed to overcome the shortcomings of human error in the aspects of comprehensiveness and consistency of error description and the effectiveness of the method itself. The current human error classification methods are divided into three categories, and the characteristics and defects of each method are analyzed. Human errors are divided into four levels, and a detailed classification framework is constructed for each level. Human errors in aviation maintenance are analyzed by this method, and the application process of the method is demonstrated. The analysis results show that the method is an effective human error classification method and can be used to guide the human error identification in various task scenarios.