Among the different negative cavitation effects that are associated with the cavitation presence (efficiency drop, noise and vibrations etc.), cavitation erosion is still one of the main limiting features in the modern trend towards designing hydraulic machines with smaller size. Cavitation erosion is particularly undesirable, because it can act in synergy with other erosion processes like corrosion and sand abrasion, which all lead to the progressive destruction of blades of runners. Although many studies attempt to predict the erosion intensity for a given set of hydrodynamic conditions, existing prediction models still need to be largely improved. Towards the development of new cavitation aggressiveness models, the context of this present work is to study the hydrodynamic mechanisms involved in cavitation erosion. The proposed approach is based on both the study of cavitation impacts and on the physical investigation of a cavitation vortex collapse. Thus, the whole study can be divided into two main parts. Firstly, a study of the cavitation aggressiveness of leading edge cavitation, which one of the most aggressive cavitation types, is achieved by performing pitting experiments in cavitation tunnel with two 2D blades of different scale. They are either placed in a rectangular test section, or in a diverging test section in order to simulate the presence of a pressure gradient. Material specimens which are exposed to a leading edge cavitation are analyzed with the help of a 3D profilometry. This study has validated the energetic model by considering the collapse efficiency which represents the global energy transfer from the vapour structures to the material. It has been shown that the flow velocity and the pressure gradient are the principal influencing macroscopic parameters for both stable and unstable leading edge cavitation. Moreover, two different kinds of cavitation attack can be suggested: the vapour structures that are created at the water-vapour interface lead to isolated impacts, the large shed transient cavities yield larger and grouped impacts. Secondly, the physical study of a cavitation vortex collapse has been carried out. This involved the cavitation vortex generator of IMHEF for the generation of an isolated vortex. The development of a visualization test section and a high-speed video shadowgraphy technique has been be required. This study has shown the complex hydrodynamic characteristics involved in the collapse of a cavitation vortex. The cavitation vortex collapse leads to an irregular and unpredictable vapour cavity shape at the final stage of the collapse. The collapse and the associated emission of shock waves are very local phenomena which can arise from different locations of the same main cavitation vortex. Estimations of the boundary velocity result in maximum values greater than 350 m s-1 at the rebound. As the collapse is not regular, local velocity at the boundary might be much higher. Furthermore, the reliability of determining the centre of collapses by using the luminescence phenomenon of the cavitation vortex is evaluated. Luminescence has been detected with the help of a photomultiplier tube and an intensified light camera. The luminescence of an isolated cavitation vortex collapse has been experimentally shown for the first time. Moreover, a simultaneous captures of luminescence and shock waves have been achieved. The luminescence of the cavitation vortex has been characterized in the terms of time occurrence, location, success rate and intensity for variable hydrodynamic conditions. The luminescence of the cavitation vortex appears as short light bursts with duration down to 10 ns and up to 100 ns. The vortex intensity is a driving parameter of the luminescence emission. Simultaneous visualizations demonstrate that luminescence sources mingle with the shock waves epicentres. Moreover, luminescence and shock wave ignition occur within a time interval which is less than 500 ns. As a remarkable result, the higher the vortex intensity is, the closer to the wall the cavitation vortex collapse takes place. Furthermore, this main result can be extrapolated to the case of transient cavitation vortices generated by a leading edge cavitation. It also explains the increase of the cavitation aggressiveness when the flow velocity increases the cavitation vortices collapse closer to the wall when the flow velocity increases.