River infiltration is an important part of groundwater recharge at riverbank filtration (RBF) sites. It is not only affected by the hydraulic gradient between the river stage and the groundwater table but also largely depends on the riverbed hydraulic conductivity (RHC). However, owing to the hydrodynamic conditions and sediment thickness during river scouring and deposition, the lithology of riverbed sediments undergoes strong spatial and temporal changes, which lead to a highly uncertain RHC. The influence of river scouring and deposition on RHC is not completely clear, resulting in calculation inaccuracies in the rate of river water infiltration to the aquifer and evaluation of the regional groundwater resource quantity. Based on the geological and hydrogeological settings at a typical RBF site in the middle reaches of the second Songhua River in China, we determined the relationship between RHC and sediment particle size using stepwise regression analysis and a genetic algorithm. We established a hydrodynamic and sediment transport model based on Delft3D to simulate the spatial distributions of sediment particle sizes. On this basis, a numerical groundwater flow model was established using Visual MODFLOW, in which the river boundary was generalized into a third type of boundary condition, and RHC zoning was performed to improve the accuracy of the simulation. We conclude that accurate prediction of RHC is essential for the assessment of groundwater resources in riparian zones. The exploitation of groundwater should fully consider the river infiltration and avoid a series of ecological and environmental problems caused by excessive groundwater exploitation. The dataset is the basic data of the above research.