The iron ores of Minas Gerais, Brazil, fall into two categories: (1) hypogene hematite ore averaging 66 percent or more Fe, and (2) lower-grade supergene ores. Most ore now extracted is high-grade hypogene ore; lower-grade supergene ores will be of much future value. All supergene ores formed by weathering of itabirite, a metamorphosed oxide-facies iron formation averaging about 38 percent Fe and 44 percent Si02. The Caue Itabirite crops out for about 540 linear kilometers in central Minas Gerais. Supergene ores fall into three intergradational categories: (1) enriched itabirite, averaging 49 percent Fe, easily con-centratable, with reserves about 25,000 million tons; (2) intermediate grade ores, averaging perhaps 63 percent Fe, with indicated and inferred reserves more than 600 million tons; and (3) canga, averaging between 57 and 62 percent Fe, with reserves in the hundreds of millions of tons. Disaggregation of hard and brittle itabirite by solution primarily of quartz and secondarily of other soluble constituents causes residual enrichment in iron with minor hydration of hematite. As weathering continues, most of the quartz is removed and more of the hematite is hydrated, producing intermediate grade ore. Secondary enrichment by limonite is important. The final weathering product is canga. In canga, almost all the iron is hydrated, and the rock is still further impoverished in Si02 and residually enriched in A1203 and P. Canga also forms by cementation of iron-rich detritus by limonite. Four factors control the supergene ore-forming process: (1) physiography, for ores occur on ridges and plateaus; (2) climate, for seasonal rainfall is apparently needed for the formation of extensive canga blankets; (3) grain size of the original itabirite, for quartz with a grain size much greater than 0.1 millimeter is not readily soluble; and (4) composition of the iron formation. The chemically inert and physically resistant canga blanket is essential to supergene ore formation because soft weathering products would otherwise be removed as fast as formed. Thus, canga gives time for the formation of other ore types. Iron fixed as cementing limonite in canga and as enriching limonite in intermediate grade ore was derived by leaching and hydration of hematite from itabirite. It probably moved in the ferrous state and was precipitated as insoluble hydrous ferric oxide primarily by oxygenation of the solutions, to a lesser extent by their evaporation at or near the surface, and to a still smaller degree by pH changes. The data show that the high-grade hematite ore, 66 percent Fe or higher, cannot have formed directly by supergene action. Geochemical processes resulting in supergene concentration of iron also concentrated alumina and phosphorus. The high-grade ore contains about the same low percentages of these materials as unweathered itabirite.