A plausible and consistent model is developed to obtain a quantitative description of the gradual disappearance of hexavalent chromium (Cr(VI)) from groundwater in a small-scale field tracer test and in batch kinetic experiments using aquifer sediments under similar chemical conditions. The data exhibit three distinct timescales. Fast reduction occurs in well-stirred batch reactors in times much less than 1 hour and is followed by slow reduction over a timescale of the order of 2 days. In the field, reduction occurs on a timescale of the order of 8 days. The model is based on the following hypotheses. The chemical reduction reaction occurs very fast, and the longer timescales are caused by diffusion resistance. Diffusion into the secondary porosity of grains causes the apparent slow reduction rate in batch experiments. In the model of the field experiments, the reducing agent, heavy Fe(II)-bearing minerals, is heterogeneously distributed in thin strata located between larger nonreducing sand lenses that comprise the bulk of the aquifer solids. It is found that reducing strata of the order of centimeters thick are sufficient to contribute enough diffusion resistance to cause the observed longest timescale in the field. A one-dimensional advection/dispersion model is formulated that describes the major experimental trends. Diffusion rates are estimated in terms of an elementary physical picture of flow through a stratified medium containing identically sized spherical grains. Both reduction and sorption reactions are included. Batch simulation results are sensitive to the fraction of reductant located at or near the surface of grains, which controls the amount of rapid reduction, and the secondary porosity, which controls the rate of slow reduction observed in batch experiments. Results of Cr(VI) transport simulations are sensitive to the thickness and relative size of the reducing stratum. Transport simulation results suggest that nearly all of the reductant must be located in the reducing stratum. Within this context and as long as there is adequate reductive capacity present, the transport simulation results are insensitive to the parameters important for the batch simulations. The results illustrate how a combination of field measurements and batch laboratory studies can be used to improve predictive modeling of contaminant transport.