Improved formulation of bedrock erosion laws requires knowledge of the actual processes operative at the bed. We present qualitative field evidence from a wide range of settings that the relative efficacy of the various processes of fluvial erosion (e.g., plucking, abrasion, cavitation, solution) is a strong function of substrate lithology, and that joint spacing, fractures, and bedding planes exert the most direct control. The relative importance of the various processes and the nature of the interplay between them are inferred from detailed observations of the morphology of erosional forms on channel bed and banks, and their spatial distributions. We find that plucking dominates wherever rocks are well jointed on a submeter scale. Hydraulic wedging of small clasts into cracks, bashing and abrasion by bedload, and chemical and physical weathering all contribute to the loosening and removal of joint blocks. In more massive rocks, abrasion by suspended sand appears to be rate limiting in the systems studied here. Concentration of erosion on downstream sides of obstacles and tight coupling between fluid-flow patterns and fine-scale morphology of erosion forms testify to the importance of abrasion by suspended-load, rather than bedload, particles. Mechanical analyses indicate that erosion by suspended-load abrasion is considerably more nonlinear in shear stress than erosion by plucking. In addition, a new analysis indicates that cavitation is more likely to occur in natural systems than previously argued. Cavitation must be considered a viable process in many actively incising bedrock channels and may contribute to the fluting and potholing of massive, unjointed rocks that is otherwise attributed to suspended-load abrasion. Direct field evidence of cavitation erosion is, however, lacking. In terms of the well-known shear-stress (or stream-power) erosion law, erosion by plucking is consistent with a slope exponent (n) of ∼2/3 to 1, whereas erosion by suspended-load abrasion is more consistent with a slope exponent of ∼5/3. Given that substrate lithology appears to dictate the dominant erosion process, this finding has important implications for long-term landscape evolution and the models used to study it.