In this paper, we consider the problem of finding itineraries in bus networks under multiple independent optimization criteria, namely arrival time at destination and number of transfers. It is also allowed to walk from one stop to another if the two stops are located within a small distance. A time–dependent model is proposed to solve this problem. While focusing on the network where the size of the Pareto set in the multi–criteria shortest path problem might grow exponentially, we develop an efficient algorithm with its speed–up techniques. An evaluation on the qualities of found paths and the empirical results of different implementations are given. The results show that the allowance of walking shortcuts between nearby stops gives a better route planning.