Transport of sulfur via mantle-derived partial melts from deep Earth to the surface reservoirs is a critical step in the deep global sulfur cycle. Given that sulfur is stored mostly in sulfide phases in mantle lithologies, the critical parameter is sulfur concentration at sulfide saturation (SCSS) of mantle-derived magmas. CO2 ± H2O-induced melting beneath oceanic and continental mantle produces incipient CO2-rich melts. Although, SCSS of silicate melts of a variety of compositions is extensively studied, the SCSS of carbonatitic and carbonated silicate melts have not received much attention. Here we present experiments in graphite capsules at pressures (P) of 2.5–6.0 GPa and temperatures (T) of 1350–1650 °C investigating the SCSS of carbonatitic and carbonated silicate melts. All experiments produced quenched Fe ± Ni-sulfide melt blobs + carbonated melt matrix ± ol ± cpx ± opx ± gt, with melt composition on a CO2-free basis varying from 7 to 40 wt.% SiO2, 0.5 to 7 wt.% Al2O3, and 9 to 17 wt.% FeO* (total FeO). SCSS measured using EPMA increases with SiO2 and T but is not affected by P; the effect of composition being more pronounced than P-T. The composition of sulfide melt phase also affects SCSS. With increasing Ni in the molten sulfide phase, the SCSS changes from 2000 to 4000 ppm (Ni-free) to is 800–3000 ppm (33 wt.% Ni). Comparison of our measured SCSS with the existing SCSS models for nominally CO2-free silicate melts and with one study for carbonated melts show that these parameterizations fail to capture the sulfide saturation values in CO2-rich melts from our study. Using our new SCSS data and previous SCSS data for melt compositions that span the range from carbonatite to basalts via carbonated silicate melts, we develop a new empirical SCSS parameterization. Unlike a previous model, which suggested SCSS of carbonated melt is only affected by melt FeO* (other than P-T) and did not constrain how SCSS evolves from low-silica carbonatitic melt to low-CO2 basaltic melt, our new parameterization captured complex effects of many melt compositional parameters, including silica on SCSS. Using our new SCSS model, we constrained the efficiency of S extraction from the mantle beneath mid-oceanic ridges and continents via low-degree carbonated melts. Deep carbonated melts beneath ridges are expected to mobilize 5–15% of the initial sulfur before nominally-volatile-free peridotite melting begins. In continental mantle, deep kimberlitic melt can act as an agent to mildly enrich the shallow mantle in sulfide as it evolves to a carbonatitic melt upon reactive cooling. Application of our data to subduction zones suggests that low degree carbonatitic melt is not an efficient agent to extract residual sulfide from the subducting oceanic crust.