Efforts to develop practical expert systems have mostly concentrated on how to implement experience-based machine learning successfully. Recently several active research projects on machine learning have been undertaken from the viewpoint of knowledge-based management. The aim of this study is to develop the Concrete Bridge Rating (Diagnosis) Prototype Expert System with machine learning, employing the combination of a neural network and bidirectional associative memories (BAM). The introduction of machine learning into this system facilitates knowledge-based refinement. By applying the system to an actual in-service bridge, it has been verified that the machine learning method employed that uses the results of questionnaire surveys involving bridge experts is effective for the system.