Banded iron formations (BIFs) have long been considered marine chemical precipitates or, as more recently proposed, the result of episodic density flows. In this study, we examined the mineralogy, petrography and chemistry of the Dales Gorge BIF to evaluate the validity of these models. Microbands reflect a compositionally variable primary precipitation/sedimentation pattern with diagenetic modifications. "Iron-rich" bands are characterized by massive anhedral aggregates and xenomorphic hematite, commonly showing overgrowths of subhedral magnetite with minor apatite and late diagenetic ankerite-Fe dolomite. "Iron-poor" bands consist of fine-grained quartz, siderite and Fe-talc in variable amounts. Amorphous silica, ferrihydrite (precursor of hematite), greenalite and possibly some siderite constitute the primary precipitates, while ankerite-Fe dolomite, Fe-talc and magnetite and the bulk of siderite are secondary mineral phases. Ankerite and Fe-dolomite most likely represent by-products of siderite dissolution and the subsequent reaction of dissolved bicarbonate with Ca, Mg and Fe. Ferroan-talc is thought to be formed from the following reactions: (i) greenalite + chert, and (ii) siderite + chert. Magnetite formed from the conversion of hematite, likely through bacterial Fe(III) reduction. Geochemical mineral analyses show that all the phases have very low concentrations of trace elements with the exception of Ba, As, Cr, Zn and Sr. This partitioning was presumably controlled by both sorptive reactions occurring during primary precipitation in the water column and secondary remobilization during diagenesis. Whole-rock analyses indicate two decoupled sources for BIF and S macrobands. While BIF macrobands have a major hydrothermal influence, data from the S macrobands supports a dominantly mafic provenance. Nonetheless, when all the lithologies (i.e., source rocks, S and BIF macrobands) are evaluated together, continuous geochemical trends can be observed. This suggests that at least part of the precursor material of BIF macrobands was sourced from the same material that gave origin to the S macrobands. Similar relationships are seen in other BIF successions, where the evolution from an ultamafic- to a mafic-dominated upper continental crust is distinctly reflected in BIF compositions through time. Interpretation of this data implies that any model developed to explain BIF deposition must consider: (i) processes involving low-temperature weathering of the continental and ocean basement rocks, mainly (ultra)mafic lithologies; and (ii) high temperature water-rock reactions associated with hydrothermal activity at spreading ridge centers or seamounts. In either case, the influence of the compositional change of the upper continental crust played a major role in the chemical compositions of BIFs through time. © 2009 Elsevier B.V. All rights reserved.