Abstract The Witwatersrand gold fields of South Africa account for more than a third of the world's total gold production since mining started there in 1886: 48,000 t Au. These gold fields dominated production throughout the twentieth century, but production peaked at 1,000 t in 1970 and has steadily declined since that time. The regional geologic framework is well understood as a consequence of intensive study and virtually unrivalled three-dimensional information derived from drilling and underground openings, supplemented by seismic data. Advancements of the last decade have come from the integration of stratigraphy, sedimentology, structural evolution, and a much greater appreciation of thermal and fluid processes. Gold has been produced from seven major gold fields located around the northern and western margin of the 350-km-long Witwatersrand basin. Each gold field consists of one major and commonly several minor reef horizons that have been mined semicontinuously for up to 400 km2. Mineralized zones range from 1 cm to several meters in thickness, and the host rock varies from quartz pebble conglomerate and carbon seam to polymictic conglomerate and pyritic quartzite. All reefs are either on or within a few meters above unconformity surfaces, and the major reefs are part of a very distinctive reef package of footwall, unconformity, conglomerate, quartz-rich sandstone and/or shale. The actual distribution of gold has been poorly represented in the literature owing to grade coding, inappropriate averaging, and omission of important features that are critical to understanding the controls on the gold. High grades and remarkable lateral continuity have nevertheless facilitated economic exploitation. The mineralogy of the Witwatersrand reefs is dominated by pyrite with lesser pyrrhotite and arsenopyrite, widespread nickel and cobalt sulfarsenides, and low base metal sulfides. A distinctive mineral assemblage of pyrophyllite-chloritoid-muscovite-chlorite-quartz-rutile-pyrite is found in and around the reef package in all gold fields. This assemblage has been used to constrain peak metamorphic temperatures to just above 300°C to the south and east of the basin, and closer to 400°C in the northwest corner west of the basin near Johannesburg. Two characteristics of the metamorphic event are the nearly strata-parallel distribution of metamorphic grade and an extremely high geothermal gradient. The presence of pyrophyllite and chloritoid in conglomerate, quartzite, shale, basalt flows, and some dikes has been used to indicate an alteration halo embracing the gold fields and approximately 300 km by 50 km (into the basin) by 3 km (of sequence). Quartz veins of up to a few centimeters thickness are common within this alteration halo especially in the reef horizons, but veins of meters thickness are rare. Ubiquitous pyrite throughout the alteration zone indicates elevated levels of sulfur in solution. The elements most closely associated with gold are Fe and C; uranium is also associated with many reefs and extensively mined as a coproduct. A consensus appears to be developing in support of a paragenetic sequence of uranium, then hydrocarbons, then gold. The origin of the Witwatersrand has been of great interest to economic geologists and a source of great contention. Historically, placer and hydrothermal theories have competed, and much more regional and mine-scale data have been used in the debate during the last decade. In the 1980s and especially the 1990s, models shifted from the unmodified to the modified placer model, and from both placer models to a hydrothermal model. Challenges still facing the placer models include regional alteration, structural control on gold, the chemical process of gold remobilization, and the poor match with modern gold placer processes. Moreover, the outstanding issues remain a source area for the gold and a convincing demonstration that processes during sedimentation concentrated gold at all. The hydrothermal replacement model invokes transport of gold into the Witwatersrand Supergroup during metamorphism and associated widespread alteration by a reduced, low-salinity fluid, analogous to other gold-only deposits. Fluids were channeled by structures, unconformity surfaces, and bedding, and gold precipitation was dominated by reaction with carbon- or iron-bearing rocks. These Fe- and C-rich rocks are concentrated immediately above unconformity surfaces. This model proposes that crustal rocks (probably mafic) beneath the Witwatersrand Supergroup are the source of gold. The hydrothermal model links the tectonic evolution of the basin to mineralizing processes, and can thus be used to target other basins with potential for similar mineralization. From the hydrothermal model arise a number of factors that contribute to the enormous size of the Witwatersrand gold fields: The age of the Witwatersrand basin, which predates a major period of gold introduction, The retroarc foreland basin, Major thrust structures around the basin margin, Unconformities in the upper Witwatersrand, Iron-rich nodules on unconformity surfaces, Carbon distribution, High geothermal gradient during gold deposition, Metamorphism and regional alteration, and Limited erosion. Special source regions for detritus and special sorting processes during sedimentation do not appear to have been critical. The Witwatersrand gold formed by the optimization of a set of processes found in other gold provinces, not by unique processes unrepresented elsewhere.