The history of atmospheric oxygen prior to the Great Oxidation Event (2.45–2.2 Ga) is not well understood. Hematite in the Marble Bar Chert from a NASA-funded drill hole (ABDP1) in the Pilbara craton, Australia, has been cited as evidence for an oxygenated ocean 3.46 b.y. ago. However, isotopic data from the same drill hole have been used to argue for an anoxic ocean. It is generally agreed that the hematite is primary, representing either a direct hydrothermal precipitate or a dehydration product of iron oxyhydroxides that formed during anoxygenic photosynthesis. Here we present new petrographic evidence from the Marble Bar Chert (in drill hole ABDP1) that shows that hematite in jasper bands formed via mineral replacement reactions. The hematite mostly occurs as sub-micron–sized inclusions within chert (so-called “dusty” hematite) that are typically arranged into polygonal clusters surrounded by a rim of clear quartz, resembling shrinkage structures. The lateral transition from laminated chert enclosing minute inclusions of greenalite, siderite, and magnetite to chert dominated by dusty hematite provides evidence for in situ replacement of iron-bearing minerals. The presence of hematite-rich bands containing octahedral crystals with residual cores of magnetite indicates that some of the hematite was derived from the replacement of magnetite. This interpretation is supported by the widespread occurrence of magnetite in jasper displaying progressive stages of replacement, from unaltered octahedral inclusions in quartz to hematite pseudomorphs along quartz grain boundaries. The occurrence of dusty hematite in fractures, sedimentary laminae, and the outer margins of polygonal clusters containing greenalite is consistent with fluid-mediated oxidation of iron-rich precursor minerals. The presence of syn-sedimentary chert breccias comprising rotated fragments of laminated chert indicates that the precursor sediment was silicified shortly after deposition. The abundance of “dusty” greenalite inclusions, which are texturally the earliest components of the laminated chert, suggests that the precursor sediment contained an iron-rich clay mineral. Our results show that hematite has replaced ferrous-rich minerals after deposition and provide a mechanism to explain the origin of hematite in the Marble Bar Chert, which is consistent with the origin of hematite in adjacent basalts. A secondary origin for hematite invalidates arguments for an oxygen-bearing ocean ∼3.46 b.y. ago and provides a viable explanation for the formation of Archean jasper bands. Our findings show that misinterpretations about the origin of hematite in early Precambrian cherts could lead to false conclusions about the chemistry of the ancient ocean and atmosphere.