Iron formations (IFs) are important geochemical repositories that provide constraints on atmospheric and ocean chemistry, prior to and during the onset of the Great Oxidation Event. Trace metal abundances and their Mo-Cr-U isotopic ratios have been widely used for investigating ocean redox processes through the Archean and Paleoproterozoic. Mineralogically, IFs consist of three main Fe-bearing fractions: (1) Fe-Ca-Mg-Mn carbonates, (2) magnetite and/or hematite and (3) Fe-silicates. These fractions are typically fine-grained on a sub-μm scale and their co-occurrence in varying amounts means that bulk-rock or microanalytical geochemical and stable isotope data can be influenced by cryptic changes in mineralogy. Fraction specific geochemical analysis has the potential to resolve mineralogical controls and reveal diagenetic versus primary precipitative controls on IF mineralogy. Here we adapt an existing sequential extraction scheme for Fe-phases (Poulton and Canfield, 2005) to the high Fe-content in IF and the specific three-fraction mineralogy. We optimized the scheme for magnetite-dominated Archean IFs using samples from the hematite-poor Asbestos Hills Subgroup IF, Transvaal Supergroup, South Africa. Previously commonly-used hydroxylamine-HCl and dithionite leaches were omitted since ferric oxides are quantitatively insignificant in these IF samples. The acetate leach was tested at variable temperatures, reaction times and under different atmospheres in order to ensure that all micro-crystalline Fe-carbonates were effectively dissolved, resulting in an optimum extraction for 48 h at 50 °C under anoxic conditions. The dissolution of magnetite by NH4-oxalate was also tested, resulting in an optimum extraction for 24 h under an ambient atmosphere. Finally, a HF-HClO4-HNO3 leach was used to dissolve the residual silicate fraction which has to date not been considered in detail in IF. Accuracy of the extraction technique was generally excellent, as verified using 1) elemental recoveries, 2) comparison of major and trace element distributions against mineralogy and 3) comparison to results from microanalytical techniques. This study focuses on the distribution of three frequently used geochemical proxies in IF; U, Mo and Cr. Molybdenum abundances in the Kuruman and Griquatown IF are low and show an apparent correlation with mineralogical variability, as determined by the sequential extraction. This suggests that changes in bulk-rock mineralogy, rather than redox chemistry might significantly affect Mo stable isotopes. For Cr, a minor bulk-rock stratigraphic increase can be related to the oxide and silicate fraction. However, a positive relationship with Zr indicates that this was also controlled by detrital or volcanic ash input. Uranium is predominantly bound to the silicate fraction and shows clear correlations with Zr and Sc implying detrital reworking under anoxic conditions. The discrepant behaviour of these three proxies indicate that mineralogy should be taken into account when interpreting heterogeneous bulk-rock samples and that fraction specific techniques will provide new insights into the evolution of atmosphere and ocean chemistry.