To better understand the Pt-As association in natural magmas, experiments were done at 1200°C and 0.1MPa to measure the solubility of Pt and Pt-arsenide phases (melt and sperrylite, PtAs2), as well as to determine the oxidation state, and identify evidence for Pt-As complexing, in molten silicate. Samples consisting of synthetic basalt contained in chromite crucibles were subject to three experimental procedures. In the first, platinum solubility in the synthetic basalt was determined without added arsenic by equilibrating the sample with a platinum source (embedded wire or bead) in a gas-mixing furnace. In the second, the sample plus a Pt-arsenide source was equilibrated in a vacuum-sealed fused quartz tube containing a solid-oxide oxygen buffer. The third approach involved two steps: first equilibrating the sample in a gas-mixing furnace, then with added arsenide melt in a sealed quartz tube. Oxygen fugacity was estimated in the latter step using chromite/melt partitioning of vanadium.Method two experiments done at high initial arsenic activity (PtAs melt+PtAs2), showed significant loss of arsenic from the sample, the result of vapour transfer to newly-formed arsenide phases in the buffer. Method three experiments showed no loss of arsenic, yielding a uniform final distribution in the sample. Analyses of run-product glasses from experiments which did not show arsenic loss reveal significant increase in arsenic concentrations with fO2, varying from ∼10ppm (FMQ-3.25) to >10,000ppm (FMQ+5.5). Despite very high arsenic loadings (>1000ppm), the solubility of Pt is similar in arsenic-bearing and arsenic-free glasses. The variation in arsenic solubility with fO2 shows a linear relationship, that when corrected for the change in the activity of dissolved arsenic with the melt ferric/ferrous ratio, yields a solubility-fO2 relationship consistent with As3+ as the dissolved species. This result is confirmed by X-ray absorption near edge structure (XANES) determination on run-product glasses.Levels of arsenic required for Pt-arsenide saturation are 50–500ppm over the fO2 range of most terrestrial basalts (FMQ to FMQ-2), >100× higher than the arsenic concentrations typical of such magmas, indicating significant enrichment of arsenic is required if Pt-arsenide saturation is to occur. In contrast, the level of dissolved Pt required to saturate in sperrylite is >8× lower than for pure Pt, suggesting that arsenic enrichment could lead to Pt removal at concentrations much less than required for pure metal saturation.