The portion of the Fe–S–O system including pyrrbotite, wüsite, magnetite, and iron has been studied between 900 and 1080 °C by modifed silica-tube techniques. At 900 °C, tie lines extend from pyrrhotite containing between 63.53 and 62.8±0.2 wt. per cent Fe to wüstite solid solution and form pyrrhotite containing between 62.8 and 60.0 wt. per cent Fe to magnetite. A ternary eutectic, troilite-wüstite-iron-liquid occurs at 915±2 °C. A ternary invariant point, where pyrrhotite (composition 62.8±0.2 wt. per cent Fe)+wüstite ⇌ magnetite+liquid occurs at 934 °C. Pyrrohotite composition strongly influences the temperature of thee magnetite-pyrrhotite solidus. Magnetite-pyrrhotite assemblages begin to melt at 934 °C when the pyrrhotite contains 62.8 wt. lper cent Fe, at 1010 °C when it contains 62.5 wt. per cent Fe, at 1030 °C when it contains 62.0 wt. per cent Fe, and at 1050 °C when it contains 60.5 wt. per cent Fe. Craig & Naldrett (1967) have shown that up to 20 wt. per cent nickel substituting for iron in pyrrhotite solid solution on a weight per cent basis has little effect on magnetite-pyrrhotite solidus temperature and that up to 2 wt. per cent copper substituting in a sunukar way lowers the solidus less than 20 °C. By redetermining the solidus in the presence of H2O at 2 kb total pressure Naldrett & Ricahrdson (1967) have show that, within experimental accuracy (± 10 °C), water has no effect on melting temperature. since natural iron-sulfide magmas rarely crystallize pyrrhotites containing more than 62.5 wt. per cent total metal, the temperature range of from 1010 to 1050 °C determined in this study is probably within 20 °C of the minimum temperature of introduction of a large number of magnatic sulfide ores. Comparison of the melting temperatures of ores with those of the rocks with which they are associated suggest that crystallization under different water pressures is responsible for the presence of sulfides disseminated as rounded ‘buck-shot’ type spherules in some rocks ans as an interstittial filling in others. The composition of an iron sufide-oxide ore magma settling from its associated silicate magma is controlled by the sulfur and oxygen fugacities of the silicate magma at the moment when equilibration between the two ceases. In the case of large bodies of massive sulfide ore, equilibration probably ceased when the ore settled out of its host; the sulfide to magnetite ratio of such ore will depen on how far below its liquidus temperature the sulfide-oxide liquid was at the moment of separation. In the case of sulfide-rich droplets remaining disseminated throughtot the plutonic host rock, equilibration probably continued to subsolidus temperatures; under these conditions it is possible that the droplets may lose all of their oxygen to the host rock. Finally in the case of sulfife-rich droplets trapped within rapidly cooled volcanic rocks complete re-equilibration was probably prevented by the rate of cooling and consequently these droplets retain much of their original oxygen as magnetite.