Railway operation management must cope with failures of the railway system or external disturbances that may cause initial delays or so-called primary delays. In heavy traffic areas of rail networks, primary delays can quickly propagate and lead to the so-called secondary or knock-on delays. This paper describes the results of experiments done to evaluate railway traffic optimization tools that enable to decrease the secondary delays by selecting appropriate route settings and sequence of the train movements. These experiments are part of a task the European FP7 project ON-TIME. The project aims to develop a prototype for a new generation of railway traffic management systems which will increase capacity and decrease delays for railway customers’ satisfaction. The results of the project will be validated through system simulation and real-life case studies proposed by railway undertakings which are partners of the project. This paper focuses on the results achieved in one of the case studies of the ON-TIME project, through an algorithm which we developed. It consists in the solution of a mixed-integer linear programming formulation for a limited computation time: the best feasible solution found within this limited computation time is the final solution returned by the algorithm. The case study tackled here represents traffic in the Gonesse junction, in France. We assess the impact of including the optimization in a rolling-horizon framework. The results show that the optimization is quite robust to different settings of the rolling-horizon framework.