The rail passenger transfer network is used to describe the bus transfer network of the peak period, on which equilibrium analysis of passenger travelling choice behavior is made based on the strategy equilibrium transit assignment. Considering the interests of passengers and the bus companies, a bi-level programming model of public transit network planning is established. Simulated annealing algorithm is designed to solve the model, through seven searching methods: deleting routes, generating routes, substituting routes, extending routes, curtailing routes, combining routes and dismantling routes to search the neighborhood solutions. The example shows that the model and algorithm are well efficient for optimizing the public transit network. ©, 2015 All right reserved.