Oxygen isotope fractionations have been determined between magnetite and water from 300 to 800°C and pressures between 10 and 215MPa. We selected three reaction pathways to investigate fractionation: (a) reaction of fine-grained magnetite with dilute aqueous NaCl solutions; (b) reduction of fine-grained hematite through reaction with dilute acetic acid; and (c) oxidation of fine iron power in either pure water or dilute NaCl solutions. Effective use of acetic acid was limited to temperatures up to about 400°C, whereas oxide-solution isotope exchange experiments were conducted at all temperatures. Equilibrium 18O/16O fractionation factors were calculated from the oxide-water experiments by means of the partial isotope exchange method, where generally four isotopically different waters were used at any given temperature. Each run product was characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), and on a limited basis, high-resolution transmission electron microscopy (HRTEM) and Mössbauer spectroscopy. Results from the microscopic examinations indicate the formation of well-crystallized octahedra and dodecahedra of magnetite where the extent of crystallization, grain size, and grain habit depend on the initial starting material, P, T, solution composition, and duration of the run. The greatest amount of oxygen isotope exchange (∼90% or greater) was observed in experiments where magnetite either recrystallized in the presence of 0.5 m NaCl from 500 to 800°C or formed from hematite reacted with 0.5 m acetic acid at 300, 350 and 400°C. Fractionation factors (103 ln αmt-H2O) determined from these partial exchange experiments exhibit a steep decrease (to more negative values) with decreasing temperature down to about 500°C, followed by shallower slope. A least-squares regression model of these partial exchange data, which accounts for analytical errors and errors generated by mass balance calculations, gives the following expression for fractionation that exhibits no minimum: 1000 ln αlmt-lw= -8.984(±0.3803)x+3.302(±0.377)x2 - 0.426(±0.092)x3 with an R2 = 0.99 for 300 ≤ T≤ 800°C (x = 106/T2). The Fe oxidation results also exhibit this type of temperature dependence but shifted to slightly more negative 103 ln α values; there is the suggestion that a kinetic isotope effect may contribute to these fractionations. A theoretical assessment of oxygen isotope fractionation using β-factors derived from heat capacity and Mössbauer temperature (second-order Doppler) shift measurements combined with known β-factors for pure water yield fractionations that are somewhat more negative compared to those determined experimentally. This deviation may be due to the combined solute effects of dissolved magnetite plus NaCl (aq), as well as an underestimation of βmt at low temperatures. The new magnetite-water experimental fractionations agree reasonably well with results reported from other experimental studies for temperatures ≥ 500°C, but differ significantly with estimates based on quasi-theoretical and empirical approaches. Calcite-magnetite and quartz-magnetite fractionation factors estimated from the combination of magnetite β's calculated in this study with those for calcite and quartz reported by Clayton and Kieffer (1991) agree very closely with experimentally determined mineral-pair fractionations.