Electric vehicles are facing a real challenge as they are expected to reduce drastically energy consumption and air pollution of transport. However their performances remain under users' expectation mainly concerning the range and its fluctuation. 'Eco-behavior' use like eco-driving could help to reduce range weakness. This paper proposes an approach based on speed profile optimization that allows the extraction of the eco-driving rules. The optimization uses an experimentally validated model associated to a dynamic programming method. The study of the more impacting parameters showed high influence of the cruising speed value whereas the acceleration rate could be considered as a second order parameter excepting for high level time constraints.