The primary objective of this study was to develop and demonstrate an experimental method to refine and better parameterize process models for reactive contaminant transport in aqueous subsurface environments and to reduce conservatism in such models without attempting to fully describe the geochemical system. Uranium was used as an example of a moderately adsorbing contaminant because of its relevance in geologic disposal of spent nuclear fuel. A fractured granodiorite from the Grimsel Test Site (GTS) in Switzerland was selected because this system has been studied extensively and field experiments have been conducted with radionuclides including uranium. We evaluated the role of pH, porous media size fraction, and flow interruptions on uranium transport. Rock cores drilled from the GTS were shipped to Los Alamos National Laboratory, characterized by x-ray diffraction and optical microscopy, and used in uranium batch sorption and column breakthrough experiments. A synthetic water was prepared that represented the porewater that would be present after groundwater interacts with bentonite backfill material near a nuclear waste package. Uranium was conservatively transported at pH8.8. Significant adsorption and subsequent desorption was observed at pH ~7, with long desorption tails resulting after switching the column injection solution to uranium-free groundwater. Our experiments were designed to better interrogate this slow desorption behavior. A three-site model predicted sorption rate constants for a pH7.2 solution with a 75-150 μm granodiorite fraction to be 3.5, 0.012, and 0.012 mL/g-h for the forward reactions and 0.49, 0.0025, and 0.001 h(-1) for the reverse reactions. Surface site densities were 1.3, 0.042, and 0.042 μmol/g for the first, second, and third sites, respectively. 10-year simulations show that including a slow binding site increases the arrival time of a uranium pulse by ~70%.