Abstract Estimates of oxygen fugacity of eclogitic rocks are linked to the redox evolution of the oceanic protolith during subduction and its residence in the lithospheric mantle, and, based on knowledge of pressures and temperatures, allow modelling of the speciation of volatile elements and diamond (or graphite) versus carbonate stability. To date, the oxygen fugacity of mantle eclogites has been shown to vary between −6 (Kasai, Congo and Udachnaya, Siberia) and −0.1 (Udachnaya, Siberia) log units (relative to the fayalite–magnetite–quartz buffer, FMQ), linked to the low Fe3+ contents of garnets. In this study, we investigated the Fe oxidation state of coexisting garnet and clinopyroxene hand-picked out of 17 diamond-free high-MgO and low-MgO mantle eclogites (dated at 2.84 Ga) from the Grib kimberlite pipe (East-European platform). Measured Fe3+/∑Fe values range between 0.03 and 0.19 for garnet and 0.18–0.38 for clinopyroxene, the former being higher than what was measured previously in garnets equilibrated at mantle conditions. The Fe3+/∑Fe of the reconstructed bulk rock ranges between 0.10 and 0.15 for high-MgO eclogites and 0.10 and 0.24 for low-MgO eclogites (with uncertainties of ± 0.02 and ± 0.03 in both cases). Thermobarometric calculations result in equilibration pressures and temperatures of 3.0–5.2 (± 0.4) GPa and 720–1050 (± 60) °C for both high-MgO and low-MgO eclogites, slightly lower than previous P–T estimates of mantle eclogites from the Udachnaya kimberlite pipe (Siberian craton). At these conditions, ∆logfo2 (FMQ) calculated using the available oxythermobarometric model varies from −1.7 to −0.6 log units for high-MgO eclogites and from −2.9 to 0.9 log units for low-MgO eclogites. Samples recording ∆logfo2 (FMQ) ≤ −1 log units overlap with North Slave, West Africa and Udachnaya eclogites, with no difference among eclogite types. The average values of −1.2 (± 0.4) log units for high-MgO and −0.6 (± 1.1) log units for low-MgO eclogites suggest different redox conditions of basaltic protoliths during subduction worldwide. Previous geochemical studies on the same rock samples reported evidence of cryptic metasomatism in both garnet and clinopyroxene that we demonstrate being not recorded by their major elements, while modal metasomatism evidenced by the presence of phlogopite as a product of interaction with a kimberlitic melt only affects the MgO of the bulk rock. Therefore, we suggest that high Fe3+/∑Fe ratios for garnet (> 0.10) and for reconstructed bulk rocks in the case of both low-MgO and high-MgO samples cannot be due to metasomatic interaction with an oxidized fluid, but rather are the consequence of Fe3+ redistribution in an unusually oxidized mafic protolith upon metamorphism. Our results highlight the redox variability of eclogites of Archaean age at conditions more oxidized than present-day mid-ocean ridge basalts (MORBs) and imply an oxidizing nature of the convective mantle source where magma was formed with consequent speciation of C in the form of carbonate fluid explaining, therefore, the lack of eclogitic diamonds in V. Grib kimberlite pipe.