Banded iron formations (BIFs) were deposited as deep-water facies distal to the late Archean Campbellrand carbonate platform, Transvaal Supergroup, South Africa. They are traditionally interpreted to have formed from iron oxides/hydroxides and silica that precipitated when upwelling hydrothermal water enriched in Fe2+ and silica mixed with cooler, shallow seawater enriched in dissolved oxygen. The ferric oxides/hydroxides were then converted to hematite or, in the presence of organic matter, reduced and incorporated into siderite. New high-resolution microscopy of BIFs from distal facies of the Campbellrand platform questions that interpretation, revealing the presence of abundant greenalite nanoparticles, interpreted to represent primary precipitates from ferruginous seawater, consistent with recent results from equivalent-aged BIFs of the Hamersley Group, Western Australia. The particles locally define primary sedimentary lamination, and are engulfed in diagenetic chert that preserves shrinkage structures, considered to have formed during dehydration and recrystallization of amorphous silica cement. The replacement of greenalite nanoparticles by coarser grained oxide, silicate and carbonate minerals, with destruction of primary textures and structures, indicates that iron-silicate muds were originally much more widespread. Based on our findings we propose a new model for the deposition of the BIFs, involving iron silicate precipitation from seawater followed by diagenetic silica cementation on the seafloor. We suggest that ferrous iron and silica were transported in hydrothermal plumes sourced from acidic vent fluids. Upon mixing with cooler, more alkaline seawater, the solubility of ferrous iron and silica fell to levels that favored rapid nucleation and the precipitation of iron-silicate nanoparticles over vast areas of the seafloor. The close association between iron and silica during the deposition of BIFs may be explained by the precipitation of iron-silicate nanoparticles and possible sorption of silica onto the surfaces of sinking nanoparticles. Furthermore, on the seafloor, iron-silicate nanoparticles presumably acted as nucleation sites for dissolved silica, promoting early diagenetic silica precipitation and cementation of the iron-rich muds. Our results indicate that iron silicates were important primary precipitates and suggest that changes in alkalinity rather than redox state played a key role in the precipitation of the precursor sediments of BIFs.