River erosion, via abrasion and plucking, plays a crucial role in the dynamics of continental landscapes. Indeed, fluvial erosion is thought to give the pace to hillslope erosion and to lead to the rapid export of produced sediments. Erosion rates and mechanisms are influenced by several factors. Among them, fractures in bedrock rivers are assumed to exert a strong control over erosion and thus, on landscape evolution. However, there is to date no systematic study of the impact of fracture geometry and density on bedrock river erosion. In this study, we investigate the impact of fracturing on erosion modes and rates of bedrock rivers using an experimental approach. The setup is an erosion mill designed to simulate the erosion of a fractured bedrock in a river. Fractured substrates are built with 3D-printed plastic (PVA) artificial fracture networks placed in concrete disks (diameter of 17 cm). To simulate erosion in a river, water and gravels covering two-thirds of the disk surface are added on the top of the disk. Water and gravels are entrained by a motor-driven propeller, inducing erosion of the disk by abrasion and plucking. We use a set of 4 cameras to monitor the disk’s topography every 2 minutes by Structure from Motion photogrammetry, allowing us to record erosion dynamics at high resolution. The impact of fracture geometry and density is explored through 36 experiments with varied fracture spacings, dips, and azimuths. Our results reveal that fracture network density influences erosion processes and the distribution of plucking and abrasion occurrence. Abrasions dominates in experiments with a a low fracture density, while experiments with a high fracture density facilitate the occurrence of plucking episodes. Fracture dip can also influence erosion processes at the scale of one disk, by generating an asymmetric network with respect to the flow direction. This tends to favour plucking on one side and abrasion on the other side of the disk. In addition, we sometimes observe spatial and temporal clustering of plucking episodes aligned with the flow direction. Finally, at the scale of the whole disk, the experimental results indicate that abrasion leads to a constant average erosion rate through time, whereas plucking induces significant spatial and temporal variations. These findings emphasize the effect of fracturing on erosion rates and modes and highlight the importance of incorporating this parameter into riverbed erosion models.