The term orogenic gold deposits has been widely accepted, but there has been continuing debate on their genesis. Early syn-sedimentary or syn-volcanic models and hydrothermal meteoric-fluid models are now invalid. Magmatic-hydrothermal models fail because of the lack of consistent spatially associated granitic intrusions and inconsistent temporal relationships. The most plausible models involve metamorphic fluids, but the source of these fluids is equivocal. Intra-basin sources within deeper segments of the hosting supracrustal successions, the underlying continental crust, subducted oceanic lithosphere with its overlying sediment wedge, and metasomatized lithosphere are all potential sources. Several features of Precambrian orogenic gold deposits are inconsistent with derivation from a continental metamorphic-fluid source. These include the presence of hypozonal deposits in amphibolite-facies domains, their anomalous multiple sulfur isotopic compositions, and problems of derivation of gold-related elements from devolatilization of dominant basalts in the sequences. The Phanerozoic deposits are largely described as hosted in greenschist-facies domains, consistent with supracrustal devolatilization models. A notable exception is the Jiaodong gold deposits of China, where ca. 120-Ma gold deposits are hosted in Precambrian crust that was metamorphosed over 2000 million years prior to gold mineralization. Other deposits in China are comparable to those in the Massif Central and elsewhere in France, in that they are hosted in amphibolite-facies domains or clearly post-date regional metamorphic events imposed on hosting supracrustal sequences. If all orogenic gold deposits have a common genesis, the only realistic source of fluid and gold is from devolatilization of a subducted oceanic slab with its overlying gold-bearing sulfide-rich sedimentary package, or the associated metasomatized mantle wedge, with CO2 released during decarbonation and S- and ore-related elements released from transformation of pyrite to pyrrhotite at about 500 °C. Although this model satisfies all geological, geochronological, isotopic, and geochemical constraints, and is consistent with limited computer-based modeling of fluid release from subduction zones, the precise mechanisms of fluid flux are model-driven and remain uncertain. From an exploration viewpoint, the model re-emphasizes the ubiquitous occurrence of orogenic gold deposits in subduction-related orogenic belts and importance of continental-scale lithosphere-tapping fault and shear zones to focus large volumes of auriferous fluid. It confirms the importance of the consistent spacing between world-class deposits, broadly equivalent to the depth of the Moho, as derived from empirical observations.