Sixty push-pull experiments were conducted to determine the factors controlling Fe(III) and Mn(IV) reduction in a well-characterized, shallow, coastal plain aquifer near Oyster, VA, USA. The five multi-level samplers each equipped with 12 ports sampled a heterogeneous portion of the aquifer from 4.4 to 8 m-bgs. Each multi-level sampler (MLS) was injected with groundwater that contained NO3- and Br- along with: (1) just groundwater (control treatment), (2) humics, (3) lactate (conducted twice) and (4) lactate plus humics. Microbially mediated Fe(III) reduction caused the aqueous FeTot concentrations to increase at every depth in the lactate treatment with significant increases within 1 day even while NO3- was present. Little change in the FeTot concentrations were observed in the control and humics treatment. Humics may have acted as an electron shuttle to increase Fe(III) reduction in the lactate plus humics treatment. The amount of Mn(IV) reduction was significantly lower than that of Fe(III) reduction. Geochemical modeling indicated that gas formation, sorption on reactive surfaces, and mineral precipitation were important processes and that Fe(III) and S O42 - reduction were co-occurring. Conditions were favorable for the precipitation of Fe-carbonates, Fe-sulfides and Fe-silicates. In the lactate treatment protist concentrations increased then decreased and planktonic cell concentrations steadily increased, whereas no change was observed in the control treatment. Correlations of Fe(III) reduction with physical and chemical heterogeneity were weak, probably as a result of the abundance of Fe(III) bearing minerals relative to electron donor abundance and that the push-pull test sampled a representative elemental volume that encompassed the microbial diversity within the aquifer. This work indicates that stimulating metal reduction in aquifer systems is a feasible method for remediating heterogeneous subsurface sites contaminated with metals and radionuclides. © 2007 Elsevier Ltd. All rights reserved.