Platinum-group element (PGE) geochemistry may be used to constrain the timing of sulfide saturation in magmas, which influences the Cu and Au fertility of evolving magmatic systems. We report new geochronological and geochemical data, with emphasis on PGE geochemistry, for a suite of regional porphyritic hornblende-diorite intrusions and ore-bearing porphyries from the super-giant Escondida and smaller Zaldivar Cu deposits of Northern Chile. The regional dioritic intrusions have zircon U-Pb ages between 396 to 371Ma, which overlap with the ages of the ore-bearing Escondida and Zaldivar porphyries (381 to 350Ma). Whole-rock major and trace element, and Sr-Nd-Pb and zircon O-Hf isotope geochemistry indicate that the regional diorites and ore-bearing porphyries are co-magmatic and originated from the same mantle-derived magma by fractional crystallization, with minor contamination by Paleozoic crust (approximate to 10%). The low concentrations of PGE in the regional diorites show that they reached sulfide saturation before the MgO content of the melt fell to 47wt %, the MgO content of the most primitive sample analysed. The fraction of sulfide melt which separated from the melts that formed the regional diorites is estimated to be approximate to 012wt %; this resulted in the partitioning of highly chalcophile elements (Au and PGE) into a sulfide phase that was retained in cumulus rocks at depth. However, the fraction of sulfide melt was too low to have a significant effect on the Cu content of the fractionating melt. As a consequence, when the evolving melt eventually reached volatile saturation, it contained enough Cu (40 +/- 10ppm) to form a super-giant Cu deposit. In contrast, Au was largely stripped from the melt by sulfide precipitation, with the result that the mineralization at Escondida is Cu dominant, with only minor Au. The Zaldivar deposit, on the other hand, contains even less Au, which is attributed to a longer fractionation interval between sulfide and volatile saturation. This study provides evidence to support previously proposed models which suggest that the timing of sulfide saturation, the amount of sulfide melt produced, the water content and oxidation state of the melt, and the magma volume are critical factors in determining the potential to form a porphyry Cu deposit. Plots of Pd/MgO against Y can be used as empirical indicators of magma fertility for porphyry mineralization, and to discriminate between Cu-Au and Cu-dominated systems, but cannot predict the size of the deposit. The super-giant status of the Escondida deposit is attributed to it being underlain by a large batholith with a calculated minimum mass of 10(12) tonnes (approximate to 400km(3)).