Iron isotopes were used to investigate iron transformation processes during an in situ field experiment for removal of dissolved Fe from reduced groundwater. This experiment provided a unique setting for exploring Fe isotope fractionation in a natural system. Oxygen-containing water was injected at a test well into an aquifer containing Fe(II)-rich reduced water, leading to oxidation of Fe(II) and precipitation of Fe(III)(hydr)oxides. Subsequently, groundwater was extracted from the same well over a time period much longer than the injection time. Since the surrounding water is rich in Fe(II), the Fe(II) concentration in the extracted water increased over time. The increase was strongly retarded in comparison to a conservative tracer added to the injected solution, indicating that adsorption of Fe(II) onto the newly formed Fe(III)(hydr)oxides occurred. A series of injection-extraction (push-pull) cycles were performed at the same well. The delta(57) Fe/(54) Fe of pre-experiment background groundwater (-0.57 +/- 0.17 parts per thousand) was lighter than the sediment leach of Fe(III) (-0.24 +/- 0.08 parts per thousand), probably due to slight fractionation (only similar to 0.3 parts per thousand) during microbial mediated reductive dissolution of Fe(III)(hydr)oxides present in the aquifer. During the experiment, Fe(II) was adsorbed from native groundwater drawn into the oxidized zone and onto Fe(III)(hydr)oxides producing a very light groundwater component with delta(57) Fe/Fe-54 as low as -4 parts per thousand, indicating that heavier Fe(II) is preferentially adsorbed to the newly formed Fe(III)(hydr)oxides surfaces. Iron concentrations increased with time of extraction, and 557 Fe/54 Fe linearly correlated with Fe concentrations (R-2 = 0.95). This pattern was reproducible over five individual cycles, indicating that the same process occurs during repeated injection/ extraction cycles. We present a reactive transport model to explain the observed abiotic fractionation due to adsorption of Fe(II) on Fe(Ill)(hydr)oxides. The fractionation is probably caused by isotopic differences in the equilibrium sorption constants of the various isotopes (K-ads and not by sorption kinetics. A fractionation factor alpha(57/54) of 1.001 fits the observed fractionation. Copyright (c) 2005 Elsevier Ltd.