A train timetable is the key to organizing railway traffic and determining the inbound and outbound times of trains. The goals for optimizing a timetable are to improve stability (the most important index) and to offer more feasibility for train rescheduling when disruptions occur, with high railway capacity utilization as a prerequisite. Train rescheduling must ensure traffic order and efficiency and adjusts train movements to be consistent with the schedule as much as possible. The two problems have the same basic solution method: to adjust the inbound and outbound times of trains at stations. Timetable stability is defined by focusing on the buffer time distribution for trains in sections and at stations. A timetable optimizing model that takes stability as the optimizing goal and a rescheduling model with minimal summary time as the destination are presented. The particle swarm algorithm is improved and is applied in problem solving as a time-adjusting tool. The algorithm is illustrated by two examples from the Guangzhou-to-Shenzhen section in China. The improved particle swarm algorithm is proved to have real-time adjusting ability, showing its high convergent speed. It is concluded that the algorithm has great global searching ability. The described new method can be embedded in the decision support tool for timetable designers and train dispatchers.