Iron oxides display a remarkable capacity for adsorbing a wide range of contaminants, including arsenic and other metals. In anoxic environments, iron oxides can undergo an abiotic reductive dissolution and/or dissolution by iron-reducing bacteria. Precise knowledge about individual processes affecting iron mobility in soils is essential to understand the release of the associated contaminants, which is of great importance for predicting the element mobility by transport modeling. Consequently, it is necessary to accurately predict iron transport before making contaminant mobility predictions. In this study, we adopt different approaches to improving consistency between actual data and model predictions for iron transport. Soil column experiments were run in duplicate under saturated flow conditions at a constant temperature of 20°C. Soil material was taken from the top horizon (A_h) of a floodplain soil near the Mulde river (Saxony-Anhalt, Germany). The redox potential and CO₂ concentration in the column effluent were recorded on a permanent basis. The pH, electric conductivity, inorganic and organic carbon, and major and trace elements were also determined. Several stop flows of various periods of time were included in the experiments. All calculations were done with the RICHY program, based on numerical modeling using the finite elements method. With either parametric or nonparametric sorption isotherms, the model predictions of iron concentrations in the effluent did not fit the obtained data. The effects of the stop flows testified that a transport model based only on thermodynamic facts did not include all the processes affecting iron mobility. Thus, the model needed refinement, especially with regard to microbial activity. Several processes that took place in the soil columns were very likely to be induced by microorganisms. First, the behaviour of the mobilised iron indicated a strong complexation by organic ligands. The effluent samples still contained ca. 25 mg L^-1 dissolved iron even after months of exposure to oxic conditions. Some Fe-reducing bacteria are known for synthesizing siderophores under conditions of iron limitation. These siderophores act as strong iron-chelating agents and are also supposed to have influence on the observed iron complexation. Second, the results of inorganic and organic carbon analyses attest that a large amount of Fe-carbonate phases was present in the effluent following the stop flows. Because the soil material was carbonate-free, the formation of Fe-carbonate phases could also be ascribed to microbial activity. Elevated concentrations of CO₂ in the effluent after flow interruptions (produced by microbial metabolism) resulted from an extended residence time in the soil column. Assuming that equilibrium conditions were present, we found that increased amounts of CO₂ led to a higher concentration of carbonate in the liquid phase. This microbial induced carbonate may have formed the presumed Fe-carbonate phase. Consequently, biotic reactions by microorganisms change the mobility of iron. Adapting the model to these conditions by considering decay as well as multiple iron species is expected to improve the agreement between model predictions and the obtained data. As a result the arsenic transport is also supposed to be predicted more precisely by the adapted model than it is by excluding single processes induced by microbial activity. This underlines the importance of identification of individual processes affecting mobility and quantifying their role in overall transport behaviour by modelling.