Although many studies have demonstrated that arc magmas are more oxidized than mid-ocean ridge (MORB) and oceanic island basalts (OIB), the oxidation state of their mantle source is still debated. This ongoing debate is mainly due to contradictory fO 2 values obtained from different proxies (e.g., Fe 3+ /ΣFe of olivine-hosted melt inclusions and glasses; Zn/ΣFe, V/Sc, V/Ga of lavas). On the one hand, some studies using V/Sc, V/Ga and Zn/ΣFe of lavas tend to show that the oxidation state of the mantle beneath arcs cannot be distinguished from that of the MORB mantle. On the other, Fe 3+ /ΣFe of glasses and olivine-hosted melt inclusions suggest that the sub-arc mantle is more oxidized than the mantle beneath ridges. Here, we estimate the oxygen fugacity of high-Mg olivine-hosted melt inclusions from various mid-ocean ridges and arcs, from one hot spot (Reunion Island) and Mount Etna using two fO 2 proxies: the Fe 3+ /ΣFe of melts and the partition coefficient of V between olivine and melt (D v Ol/Melt). After assessing the role of secondary processes such as volatile degassing and fractional crystallization on the fO 2 of melts and reconstructing primary melt compositions, we show that (1) fO 2 values derived from Fe 3+ /ΣFe and D v Ol/Melt are comparable and (2) arc and Mount Etna primary melts are more oxidized than mid-ocean ridge and Reunion Island primary melts. We then demonstrate, from Zr/Nb, that the observed variability in primary melt fO 2 is not due to chemical variability of the mantle source. Finally, the correlations between incompatible trace element ratios such as Th/La, Ba/La, Ba/Th and La/Yb and the fO 2 of primary melts reveal a link between the oxidized nature of arc and Mount Etna primary magmas to slab fluid and/or sediment melt influence. Our arc dataset displays a variety of subduction influences, from fluid-dominated (Aoba and Mount Meager) to sediment melt-dominated (La Sommata) influences. The origin of the oxidation of Mount Etna magmas is more complicated to identify and the nature of the oxidized metasomatic fluids that likely percolated through the mantle source before magma generation is yet to be determined.