Within diesel engines, the valve-seat contact is one of the few non-lubricated contacts which is subjected to significant degradation. This degradation is put in evidence by material removal at the intake valve. The material pull out is promoted by the replication of combustion cycles (500 million) and severe operating conditions (pressure 18 MPa). The wear can lead to gas leakage and engine failure. The target of this work was to identify the main parameters affecting this wear. Our approach was based on the tribological triplet and material flows within the contact involving both numerical and experimental approaches. A dynamic model and a valvetrain test bench showed that the wear flows could be activated by the architecture of the valve opening system. Consequently, the limitation of these flows can be obtained by controlling the “global” geometry of the system and therefore without modifying the properties of the materials. In the same way, a finite element model of the local response of the seat-valve contact highlighted the impact of the “local” geometry of the contact. The change of this geometry is a lever to limit the shearing forces which reduces the tearing of the particles and therefore wear. Finally, tests carried out on the engine and on a specifically adapted test bench completed the understanding of degradation mechanisms (source flow, wear flow, etc.). Morphological interpretations of worn surfaces in terms of material flows allowed the understanding of the build-up stages of a protective layer. One solution to promote this internal flow is the use of pollutants from combustion. For example, the burned oil in contact, which is a priori harmful, becomes an opportunity here. In addition, un-burned hydrocarbons from the combustion of biodiesel help to protect the contact.