We examine controls on mantle oxygen fugacity (fO 2 ) during the partial melting process that forms mantle lithosphere at spreading centers. We compare the paleo fO 2 at the time of melting inferred by V/Sc systematics of ophiolite peridotites, with the thermobarometric fO 2 recorded by olivine–orthopyroxene–spinel assemblages during simple cooling in relatively young oceanic lithosphere. Modelling of the V/Sc in the ophiolite peridotites from Alaska, Yukon and British Columbia is permissive of only a narrow range in log fO 2 during melting between NNO and NNO−1 (where NNO is the nickel–nickel oxide buffer), depending on the choice of partition coefficients for Sc. This result is within uncertainty of the thermobarometric fO 2 recorded by most samples (within 1 log unit of NNO−1). The same cannot be said for more complex peridotite residues from continental mantle, where V/Sc systematics show a narrow paleo fO 2 during formation but wide range of thermobarometric fO 2 after equilibration in the lithosphere. In continental mantle with a complex history, thermobarometric fO 2 is an ambiguous measure of that attendant during partial melting. Graphite-saturated melting in a system closed to oxygen controls melt Fe 3+ /Fe 2+ and CO 2 content, and creates a shift in fO 2 of about 2 log units [1] in a peridotite residue. In contrast, for the ophiolite mantle samples in this study, both paleo- and thermobarometric fO 2 are near values predicted by carbon–fluid equilibria, yet show no relationship with depletion, suggesting the melt-residue system in the mantle may be open to oxygen during the partial melting process.