This paper compares the French and Japanese scour risk assessment procedures, namely the machine learning (ML) model from the French National Railway Company (SNCF) and the scoring table proposed by the Railway Technical Research Institute (RTRI) in Japan. To demonstrate how to use the guidelines and make a comparison, they are applied to two bridges located in Japan and France respectively. In general, both guidelines aim to help screen high scour risk structures in an effective way. However, due to the different modes of inspection and hydrologic conditions, the parameters required in each approach are quite different. The two approaches are detailly compared and future directions for improvement are proposed. Results of this paper could be served as an insightful reference to institutes or countries who want to develop their own guidelines.