Most marine accidents related to vessels are the result of human factors, such as missed timely detection, judgment errors and incorrect operation of functional systems for safe navigation. In recent years, collision accidents are occurring frequently in the seas off the coast of Japan, which have been reported widely in the media. These accidents convey a negative image of marine transportation and needs to be addressed for improvements. In this paper, we examine the application of the Event Tree Analysis model and the Bayesian network model for accident occurrence modeling, which can reveal the causal relationships between human factors in maneuvering vessels and ship collisions. Then we create risk analysis models for the collision accidents based on Formal Safety Assessment (FSA) approved by IMO in 2002. These are applied by analyzing the records of adjudication of marine accident inquiries of ship collisions between fishing vessels and cargo vessels mainly.