In this paper an approach is proposed for optimization of speed profiles in railway integrating energy saving. This approach deals with a multi-objective problem involving three criteria: reduction of the travel duration, reduction of the delays and minimization of the energy consumption. Based on a state of the art evolutionary algorithm, the proposed approach searches for diversified solutions in a continuous search space. These solutions are evaluated and compared according to the Pareto approach in such a way that the proposed solutions are different and incomparable, in order to help the decision makers. After having reminded railway dynamics elements, the approach is detailed as well as the evolutionary algorithm and its problem-related components. Two case studies (the Gonesse junction and the line Saint- ´ Etienne – Rive de Giers in France) and results are then provided and analyzed.