In this experiment, 28 participants completed an urban driving task in a highly immersive driving simulator at Renault’s Technical Centre for Simulation. This simulator provides a 150 deg field of view in a fully instrumented cockpit. Two different eco-driving assistance devices were added: a visual display on the midconsole and a force feedback system on the gas pedal, in order to apply an additionnal reaction torque on drivers’ foot. The feedback information was computed by comparing the car’s instant acceleration with an optimal acceleration level based on a proprietary consumption model of a Renault diesel engine. This experiment has three main goals: I. Assess the contribution of verbal instructions to eco-driving performance; II. Quantify the additional contribution generated by two eco-driving assistance systems (visual and haptic); III. Measure drivers’ acceptance of haptic eco-driving assistance system. Basic eco-driving instructions, such as changing gears under 2000 Rpm, yield significant decrease of polluting emissions. Assisting drivers with visual, haptic, or visual-haptic on-board devices, in addition to low engine speed verbal instructions, lead to supplementary significant savings of polluting emissions. There is no significant difference between assistance feedback type; suggesting that the haptic feedback provides the same ecoperformance as visual feedback. In particular, subjects show good adaptation to the haptic feedback pedal at first utilization of the system. They apparently relied more on haptic modality to achieve the eco-driving task, when they used both visual and haptic assistance.