Fault tree, faulty petri net and Bayesian network are three powerful models used to assess a system’s reliability and safety. Fault tree is the most widely used while each of the latter two has its own irreplaceable merits, which can make up fault tree’s deficiency in practice. Based on the basic knowledge about these three models, both method for mapping fault trees into faulty petri nets and method for mapping fault trees into Bayesian networks are discussed. Combing with these two methods stated above, a table that shows these three models’ equivalent classic logic topology is presented, which reveals the inner connection among these three models. On the basis of the table, a method for mapping a faulty petri net into Bayesian network is raised. Meanwhile, a failure event of unexpected ignition of an engine is studied and compared using these three models separately and the results show that faulty petri net describes the process of fault propagation clear and using the equivalent Bayesian network that mapped from a petri net or fault tree, more quantitative information can be calculated.