The measurement of cosmogenic nuclide (CN) concentrations in riverine sediment has provided breakthroughs in our understanding of landscape evolution. Yet, the link between this detrital CN signal and landscape evolution is based on hypotheses that are not easy to verify in the field. New applications could arise from a better understanding of the statistics of CN concentrations in sediment grains. In this work, we present the coupling between the landscape evolution model Cidre and a model of the CN concentration in distinct grains. These grains are exhumed and detached from the bedrock and then transported in the sediment to the catchment outlet with temporary burials and travels according to the erosion-deposition rates calculated spatially in Cidre. The concentration in the various CN can be monitored in these grains. Because the CN concentrations are calculated in a limited number of grains, they provide an approximation of the whole CN flux. Thus, this approach is limited by the number of grains that can be handled in a reasonable computing time. On the other hand, part of the variability in the erosion-deposition processes can be recorded in the grain-by-grain distribution of the CN concentrations by monitoring the CN concentrations in distinct grains using a Lagrangian approach. We illustrate the robustness, the perspective and the limits of this approach by deriving the catchment-mean erosion from the 10Be mean concentration of the grains leaving a synthetic catchment uplifting at different rates and by comparing this derived erosion rate to the actual one calculated by Cidre.