The automotive industry and the design of engines are strongly ruled by performance and legislation demands. First order aspects defining the main functions of a component are well known and new challenges appear at a lower scale. Wear reduction is defined as a second order problem and requires decreasing the scale of observation of components down to roughness. This thesis work and its aim lie within the scope of wear improvements and are focused on the injection cam/roller contact of heavy duty diesel engines. In order to reach the aim several material and methods are used: a valve train rig test, roughness characterisation and different manufacturing processes. It is shown by experiments that the flank and nose of injection cams are specific areas with specific roughness (higher density of summits Sds) compared to the rest of the cam. The wear is characterized by rounded-off of summits (Ssc). From those results and since wear processes are strongly linked to the pressure generated between the mating surfaces, the development of a rough contact model is of great interest. An elastic rough contact model is implemented and improved later on by an elasto-plastic description of materials. The simulations are validated by a wear test and are used to rank the ability of surfaces (both cam and roller) to face wear problems. The ratio of plastically deformed peaks shows that the flank and nose of the cam are reacting badly to pressures. Additionally the rough contact allows ranking/optimization of different machining processes with respect to their expected functional contact performance. Future work will be to use such a model to choose a good combination of surfaces (cam and roller) in order to reduce wear.