The Eocene-Miocene volcanic and hypabyssal rocks of the El Indio-Pascua Au-Ag-Cu belt in the southern central Andean flat-slab region are medium-high-K calc-alkaline arc suites, ranging in composition from andesite to rhyolite. A significant transition in magmatic trace element chemistry, coinciding with a pronounced reduction in magma output, occurred in the late-Middle Miocene as documented by ⁴⁰Ar-³⁹Ar geochronology. The upper Eocene-lower-Middle Miocene rocks exhibit low Sr/Y ratios (<50), minor heavy REE fractionation with Sm/Yb ratios not exceeding 3.5 and, in some cases, minor negative Eu anomalies. In contrast, the largely dacitic rocks erupted after ca. 13 Ma are depleted in Y (± 10 ppm), have generally high, but variable Sr/Y ratios (30-200), exhibit moderate middle and heavy REE fractionation (Sm/Yb: 3.7-5.9) and lack negative Eu anomalies. The latter features are characteristic of adakitic suites (i.e. slab-melts), but the regional temporal and spatial distribution of arc magmatism precludes a major magma source in the downgoing slab. This evolution is interpreted as reflecting a progressive increase in pressure and the availability of water in the lower-crustal site of magma generation, establishing both garnet and hornblende as major stable phases in the residuum. The pressure in the lower crust increased in response to episodic crustal thickening related to the shallowing of the slab, a process recorded by the incision of three regional pediplains over the period 17-6 Ma. Elimination of the subarc asthenospheric mantle and much of the lithospheric mantle by ca. 10 Ma permitted direct incursion of slab-derived, highly oxidised metal- and volatile-rich super-critical fluids into the lower crust, stimulating melting of mafic, garnet amphibolitic and eclogitic assemblages. The igneous suites emplaced from 36-11 Ma were associated with widespread, and locally intense, epizonal hydrothermal activity, but this was barren of base and precious metals. The shallow-crustal availability of abundant water highlighted in earlier models was therefore not a metallogenetic determinant. Moreover, economic Au-Ag-Cu mineralization, associated with small volumes of dacitic magma, was restricted to the interval 9.5-5 Ma, and was not initiated until at least 3.5 my after the inception of high-pressure magma generation. In contrast to previous metallogenetic studies, we therefore suggest that this petrochemical transition was not inherently favourable for ore formation. We propose that the incursion of highly oxidized supercritical fluids from the slab into the lower crust was ultimately responsible for the brief Late Miocene metallogenetic episode.