On commuter lines, trains are often forced to decelerate or stop between stations as a result of delays of the preceding train. Once a train stops between stations, both the headway and the travel time increase. To solve these problems, we propose a train control method based on the prediction of train movement. We estimated the effects of this method on the time taken to recover from train delays using a model of the recovery mechanism, and evaluated its effect on train traffic by computer simulation. The results confirmed that the proposed method was effective in reducing the amount of time trains stopped between stations as well as recovery time.