The giant Bingham Canyon porphyry Cu-Mo-Au deposit (Utah) is associated with Eocene subvolcanic intrusions. It shows a distinct metal zonation above a barren core, with dominantly shallow Cu-Au mineralization (Cu stage) following the early quartz monzonite porphyry (QMP) intrusion, and spatially deeper Mo mineralization (Mo stage) occurring in a separate vein set exclusively after a late quartz latite porphyry (QLP) intrusion that truncates earlier Cu-Au veins. To understand this metal separation and the geochemical process of molybdenite mineralization, we investigated fluid inclusions by microthermometry, Raman spectroscopy, and laser ablation inductively couple plasma mass spectrometry (LA-ICP-MS) microanalysis in low- and high-grade quartz veins of both mineralization stages. In deep, low-grade quartz veins interpreted to represent the root zone of the Cu stage we found high concentrations of Cu, S, and Mo in the fluid inclusions, whereas in low-grade Mo-stage veins, we found lower Cu, but similar concentrations of S and Mo, compared to the inferred input fluids to the Cu stage. Sulfur and copper concentrations were similar in intermediate-density-type fluid inclusions in deep low-grade Cu-stage samples, whereas intermediate-density-type inclusions in low-grade Mo-stage veins have S contents that exceed their Cu contents. In high-grade Mo-stage vein, we found large variations of Mo concentrations in coexisting brine and vapor inclusions. Compared to the P-T conditions of the Cu precipitation stage (90–260 bars and 320°–430°C), the Mo-precipitating fluids were trapped at higher pressures and temperatures of 140 to 710 bars and 360° to 580°C. Mass-balance calculation based on the compositions of intermediate-density inclusions and brine + vapor assemblages, interpreted to be derived by phase separation during decompression of the ascending single-phase intermediate-density fluid, indicate that the mass of vapor phase exceeded that of brine by about 9:1 in both mineralization stages. Combining this mass balance with the analyzed vapor/brine partitioning data indicates that more than 70% of Mo and S (by mass) in the deposit were deposited from the vapor phase. Earlier Cu-Au deposition was similarly dominated by vapor, but recently published data about postentrapment Cu diffusion in and out of fluid inclusions cast doubt on previous quantifications, suggesting that almost none of the copper was deposited by brine. Mo is less likely to be modified by selective diffusion, and high Mo contents (max 0.0054 Mo/Na in intermediate density; 380 μ g/g Mo in brine) in the hydrothermal fluids were maintained from the early Cu stage to the late Mo stage. This indicates that Mo concentration was not the decisive factor for separate precipitation of late Mo ore at Bingham Canyon. Instead, the metal separation may be explained by a reduction in redox potential and an increase in acidity in the evolving source region of the fluids, i.e., a large subvolcanic magma reservoir. This is indicated by the stoichiometry of chalcopyrite and molybdenite precipitation reactions, a tentative difference in the Fe/Mn ratio in fluids of both veining stages, incipient muscovite alteration along high-temperature molybdenite veins, and an increasing tendency for Mo to fractionate from brine to vapor. We suggest that the early Cu-stage fluids were slightly more oxidized and neutral, allowing Cu-Fe sulfides to saturate first, while molybdenite saturation was suppressed and Mo was lost from the early ore stage. By contrast during the later Mo stage, the fluids were more reduced and acidic, thereby allowing selective saturation of molybdenite as the first precipitating sulfide in the cooling and expanding two-phase fluid, consistent with textural observations. This interpretation may imply more generally that small differences in redox potential and acid/base balance of the magmatic source of porphyry-mineralizing systems may be decisive in the temporal and spatial separation of the two metals.