High-grade BIF-hosted iron ore deposits are widely believed to have formed by epigenetic residual enrichment of hematite at the expense of other constituents, most notably chert. Processes responsible for the enrichment to high-grade iron ores are, however, only poorly understood and a range of metallogenetic models have been proposed. Field relationships have been used to distinguish three major groups of BIF-hosted high-grade iron ore deposits, namely deposits of ancient supergene, hydrothermal, and supergene-modified hydrothermal origin. Iron ores from all three deposit types are essentially composed of hematite; among different morphological types of hematite, microcystalline platy hematite and martite predominate. In this contribution, the oxygen isotope geochemistry of ore-forming hematite and martite from several high-grade iron ore deposits is examined, in an attempt to differentiate deposits of hydrothermal origin from those formed in ancient supergene environments. The δ 18 O composition of martite and microplaty hematite from deposits presumed to be of hydrothermal origin range from +0.9‰ to −7.3‰. Microcrystalline platy hematite from high-grade ores of ancient supergene origin, in contrast, has δ 18 O values ranging between +2.0‰ and −3.9‰. The latter range overlaps with the range that is defined by hematite and magnetite from weakly metamorphosed Archean–Paleoproterozoic BIF (+5‰ to −4‰). The results obtained for ancient supergene deposits developed along the 2.2 Ga Gamagara-Mapedi unconformity strengthen the argument that the Paleoproterozoic atmosphere-hydrosphere-lithosphere system was very similar to that of modern Earth. The marked shift to negative δ 18 O values displayed by hematite and martite from hydrothermal iron ore deposits, on the other hand, provides support for the suggestion that aqueous fluids of shallow crustal origin were responsible for the hydrothermal enrichment of banded iron formations to high-grade iron ores.