Accurate estimates of the sulfur content in silicate melts at sulfide saturation are required to explain magmatic processes such as the origin and budget of sulfur in some volcanic eruptions and the transfer of Cu, Au, and other precious metals from the mantle to the crust via partial melting of the mantle. A model that links S speciation, S content at sulfide saturation (SCSS), S content at anhydrite saturation (SCAS), and changes in oxygen fugacity (fO2) is presented. The model predicts an exponential increase in the SCSS with increasing fO2 from fayalite-magnetite-quartz (FMQ) because of the contribution of sulfate species in the melt. The predicted SCSS ranges from 1300 ppm (at FMQ − 1) to 1500 ppm (at FMQ + 0.5) for mid-oceanic-ridge basalts (MORB), is up to 7500 ppm (at FMQ + 2) for backarc and oceanic-island basalts, and can be as high as 1.4 wt% (at FMQ + 2.3) for island arc basalts. For the mantle wedge above subducting slabs, the SCSS ranges from 1500 ppm (at FMQ + 0.4) to 4500 ppm (at FMQ + 1.7). Disequilibrium experiments show textural evidence that is consistent with the model predictions. Thus, sulfide-undersaturated magmas can be formed at relatively low degrees of melting in metasomatized mantle sources without the need to invoke sulfide elimination by oxidation at high fO2. Estimates of the amount of S that can be transported by silicate melts need to account for the enhanced capacity of silicate melts to dissolve S when sulfate species are present, even if sulfides are the stable S-bearing phase.