Abstract: Bayesian Network(BN) has advantages in dealing with the large scale complex system and can be used to undertake safety analysis of marine boiler system.Attempts have been made to conduct such analysis by combining the BN theory with the Fault Tree Analysis(FTA) method.In this study,particular attention was given to the automatic updating ability of the BN method as well as its capability of reverse reasoning from posteriori in nformation to apriori information.Using this methods,the whole system was updated automatically by upgrading the FT status of detection units and therefore the events of maximum probability were determined.The results indicate that using this method can make the safety assessment more accurately and rapidly and also have the potential to improve the analysis＇s efficiency and greatly mitigate the risks brought by the accidents.