Recently, while studying the oxygen reduction reaction (ORR) in alkaline medium on Mn oxides with various crystal structures, we documented [1] an unparalleled activity of Mn2O3 oxide with bixbyite structure. Then we also observed a clear correlation of the electrocatalytic activity with the formal potential of Mn(IV/III) interfacial redox transition and suggested the ORR mechanism involving this transition as a mediator step. To extend our understanding of this complex mechanism, we also compared the behavior of various oxides in peroxide oxidation/reduction reactions [2] and in respect to the peroxide yield during the ORR. All materials were studied in the form of composites with Sibunit carbon (weight ratio 1:1). This carbon binder provides a high utilization of the oxide surface in electrocatalysis [3]. The voltammetry in deairated 1M NaOH solution and rotating disc electrode (RDE) measurements in the same but O2-saturated solution were used to determine the total charge and the ORR kinetic currents. Rotating ring disc electrode (RRDE) was applied to compare the peroxide yields on various Mn oxides. All the values were compared after normalizing them to the BET surface area. The number of electrons involved in the oxygen reduction was close to 4 and 3 for more and less active oxides, respectively. Less active oxides demonstrated significantly higher peroxide yields and slower peroxide reduction, with a convection-independent limiting current which documents a slow chemical step in a wide potential region. The independent experimental facts which can be considered for comparison with simulations are as follows: surface concentration of the Mn active centers (estimated from the total charge in Mn(IV/III) transition region); specific activity of pure carbon generating additional peroxide, which further reacts on the oxide component in carbon-oxide composites (see in Ref.[4] about the dual role of carbon); ORR polarization curves; polarization curves for peroxide reactions; peroxide yield during the ORR. This wide set of data provides a valuable input for multi-parametric simulations, which significantly decreases the number of possible solutions. The modeling involved the rate constants for five reaction steps as well as the parameter for interactions of the adsorbed oxygen species involved in the Mn(IV/III) transition (in terms of Frumkin isotherm). We found that the complete set of data can be interpreted only assuming the structural dependence of at least two rate constants: one at the initial stage (peroxide formation) and another at the final stage (peroxide reduction). Another intriguing possibility to explain the very high Mn2O3 activity in the ORR is to assume a direct ORR pathway for this oxide, and to consider a series pathway for all other oxides. To elucidate whether the latter hypothesis is reasonable, we addressed the O-O bond break on Mn2O3 and MnOOH (typical examples of high and low activities, respectively) in the framework of a quantum chemical approach resting on the cluster model of the oxide surface and the DFT level of theory. The computational results were used to map the effective reaction path. We found that for both model surfaces the direct bond cleavage at the oxide surface is hardly feasable both for the O2 molecule and most likely for peroxide intermediates. This finding makes it possible to exclude the direct pathway and to focus on molecular reasons of the structural effects on the rate constant for other steps such as the reduction of the adsorbed peroxide intermediate. The work is supported by EraNetRus program, project (#270 NANO-Morf). [1] A.S. Ryabova et al, Electrochimica Acta 187 (2016) 161–172. [2] A.S. Ryabova et al, ChemElectroChem 3 (2016) 1667 – 1677. [3] A.S. Ryabova et al, Electrochimica Acta 246 (2017) 643–653. [4] T. Poux et al, Catalysis Today 189 (2012) 83– 92.