The ability of reactive transport models to accurately replicate the fate of contaminants in SUDS has been sparsely documented. The present work aims (i) to improve the description of the processes, including non-uniform fluxes at the surface, and (ii) to use extensive experimental data, collected in a stormwater infiltration basin, to evaluate the model's accuracy. Both the model and observations confirm that most infiltration occurs near the water inlet. Differences between the measured and predicted profiles are either attributable to the filtration of particle-bound metals, or to spatially variable sorption capacities of the soil – notably because of lower contents of organic matter.