It has recently been proposed that the Fe3O4 content of spinel coexisting with olivine and orthopyroxene may be used as an oxygen barometer for spinel peridotites (O'Neill and Wall, 1987; Mattioli and Wood,1988). This barometer relies on the equilibrium: 6 Fe2SiO4 (olivine) +O2= 3 Fe2SiO6 (orthopyroxene) +2Fe3O4 (spinel). The two calibrations, both based ona regular solution model for olivine (WFeMg = 8.36 kJ/mol),ideal two-site model for orthopyroxene, and different activity data for spinel,generally give results within 0.7 log fo2 units of each other for mantle compositions. These calibrations have been tested, together with arecent spinel model of Nell and Wood (1990) by reequilibrating mixtures of olivine (Fo89) orthopyroxene (En90) and spinel of varying composition at known T and fo2 at 1 atm pressure. Experiments were performed at 1200°C and oxygen fugacities between fayalite-Magnetite-quartz buffer (FMQ) and 1.4 log units below FMQ. The equilibrium Fe3O4 content of spinel was reversed in experiments of 5–7 days duration at ratios ofCr/Cr+Al (spinel) of 0.13–0.53. Phases were analyzed by electron microprobewith Fe3+/total Fe of spinel being determined using spinelmicroprobe standards of known ferric iron content. The oxygen barometer wasthen tested by comparing oxygen fugacities calculated from the compositions ofexperimental products with the actual fo2 of the experiment.The Mattioli-Wood version of the barometer gives fo2 values which average close (−0.15 log units) to the true values but which show a dependence on (Cr/Cr+Al) in the spinel. The O'Neill-Wall version gives average values which are 0.65 log units low but which show no compositional bias. TheNell-Wood version gives composition-independent values which are closest to thetrue values, averaging 0.35 log units low. All three versions are sufficiently accurate to be petrologically useful and can be corrected, using these experimental data, to approximate the true values within a few tenths of a log unit in fo2.