Intermediate-scale experiments (meter-long, two-dimensional flow cell) were performed with aerobic biodegradation of benzoate substrate in physically heterogeneous (bimodal inclusive) media. Clastic heterogeneities were represented in a quasi-two-dimensional field, with low-conductivity inclusions embedded in a high-conductivity sandy matrix. The two media had similar pore-scale dispersivities but the conductivity ratio (~1:50) incurred macrodispersive spreading in the longitudinal direction. The high-conductivity sand was uniformly inoculated with Pseudomonas cepacia sp., and a pulse input of substrate and chloride ion tracer were evaluated. Degradation and growth were oxygen-limited under nonlinear dual-Monod kinetics and controlled by spatial and temporal variations in nutrient flux. The low-conductivity inclusions created regions of slow transport that prolonged the dual availability of both oxygen and substrate, which in turn enhanced microbial growth in these regions. Bacterial detachment was significant, and the fivefold increase in biomass due to growth was entirely accounted for in the aqueous effluent which displayed a complicated nonlinear breakthrough curve. High-resolution deterministic modeling was applied to simulate the intermediate-scale experiment, with parameters of the relevant constitutive relations calibrated independently through batch and small-scale column experiments. Parameter fitting to match flow cell data was avoided. This approach was taken in order both to test the predictive modeling capability as it would necessarily be used in a field application and to avoid the a priori assumption that all relevant processes were adequately represented in the respective constitutive theories. Analyses of the fit between the independently calibrated model and the flow cell data were then used to isolate processes for further experimental study. This iterative experimental/modeling approach identified processes that contributed (surprisingly) to biodegradation in heterogeneous media and yet are not currently incorporated in most mathematical models: (1) buoyancy effects associated with very small solution density variations, amplified in heterogeneous media, and (2) dynamic biological processes associated with growth, namely, endogenous respiration, cell division partitioning to the aqueous phase, and active adhesion/detachment that are strongly coupled to the transport of dissolved nutrients or microorganisms.¿ 1997 American Geophysical Union