Volcanic gases collected during episode 1 of the Puu Oo eruption along the east rift zone of Kilauea Volcano, Hawaii, have uniform C-O-H-S-Cl-F compositions that are sharply depleted in CO₂. The CO₂-poor gases are typical of Type II volcanic gases (gerlach and Graeber, 1985) and were emitted from evolved magma stored for a prolonged period of time in the east rift zone after releasing CO₂-rich gases during an earlier period of temporary residence in the summit magma chamber. The samples are remarkably free of contamination by atmospheric gases and meteoric water. Thermodynamic evaluation of the analytical data shows that the episode 1 gases have equilibrium compositions appropriate for temperatures between 935 and 1032°C. Open- and closed-system equilibrium models of species distributions for the episode 1 gases show unequivocally that coexisting lavas buffered the gas oxygen fugacities during cooling. These models indicate that the f_O2 buffering process occurs by transfer of oxygen from the major species in the gas phase (H₂O, CO₂, SO₂) to the lava during cooling and that the transfer of oxygen also controls the fugacities of several minor and trace species (H₂, CO, H₂S, S₂, Cl₂, F₂), in addition to O₂ during cooling. Gas/lava exchanges of other components are apparently insignificant and exert little influence, compared to oxygen exchange, during cooling. Oxygen transfer during cooling is variable, presumably reflecting short-term fluctuations in gas flow rates. Higher flow rates restrict the time available for gas/lava oxygen transfer and result in gases with higher equilibrium temperatures. Lower flow rates favor f_O2-constrained equilibration by oxygen transfer down to lower temperatures. Thus, the chemical equilibrium preserved in these gases is a heterogeneous equilibrium constrained by oxygen fugacity, and the equilibrium temperatures implied by the compositions of the gases reflect the temperatures at which gas/lava oxygen exchange ceased. This conclusion challenges the common assumption that volcanic gases are released from lava in a state of chemical equilibrium and then continue equilibrating homogeneously with falling temperature until reaction rates are unable to keep pace with cooling. No evidence is found, moreover, that certain gas species are kinetically more responsive and able to equilibrate down to lower temperatures than those of the last gas/lava oxygen exchange. Homogeneous reaction rates in the gas phase are apparently slow compared to the time it took for the gases to move from the last site of gas/lava equilibration to the site of collection. An earlier set of data for higher temperature CO₂-rich Type I volcanic gases, which come from sustained summit lava lake eruptions supplied by magma that experienced substantially shorter periods of crustal storage, shows f_O2 buffering by oxygen transfer up to 1185°C. Oxygen fugacity measurements in drill holes into ponded lava flows suggest that buffering by oxygen transfer may control the f_O2 of residual gases down to several hundred degrees below the solidus in the early stages of cooling. Although the details of the f_O2 buffering mechanisms for oxygen transfer are unknown, the fact that f_O2 buffering is effective from molten to subsolidus conditions suggests that the reaction mechanisms must change with cooling as the reactants change from predominantly melt, to melt plus crystals, to glass plus crystals. Mass balance calculations suggest that redox reactions between the gas and ferrous/ferric iron in the lava are plausible mechanisms for the oxygen transfer and that the f_O2 of the gases is buffered by sliding ferrous/ferric equilibria in the erupting lavas. Contrary to expectations based on models predicting the oxidation of basalt by H₂ and CO escape during crustal storage, CO₂-rich Type I gases and CO₂-poor Type II gases have identical oxygen fugacities despite greatly different crustal storage and degassing histories. Volcanic gas data give a tightly constrained log f_O2 of NNO − 0.5 (±0.05) for subaerially erupted Kilauea basalt from liquidus to solidus temperatures, consistent with recent f_O2 determinations for the mantle source regions of ocean island basalts. Because the oxygen fugacities of volcanic gases emitted by subaerial lavas imply that the f_O2 of Kilauea basalt is unchanged during crustal storage, Kilauea basalt either arrives in the crust with an oxygen fugacity between NNO and FMQ, or it develops an oxygen fugacity in this range immediately upon arrival in the summit chamber.