Electrocatalysts for the oxygen reduction and the oxygen evolution reactions play a key role and determine the power and efficiency of energy conversion/storage devices, such as metal–air batteries or regenerative fuel cells. Metal–air batteries are gaining increased interest given their high power and energy density, versatility and reversibility. The air electrode catalysts must be bi-functional and highly active, being the reactions involved at this electrode critical for the overall efficiency of metal–air batteries. Oxygen electro-oxidation/electro-reduction presents slow kinetics, thus, in order to achieve high current densities, porous high surface area electrodes are required. Carbon materials are commonly used to increase the electrode surface area. The main disadvantage is that carbon corrosion processes can occur during the battery charging process, where oxygen is evolved at very positive potentials, thus affecting the stability of the catalysts. One way to reduce the carbon corrosion rate is the use of graphitized carbon materials. In the present work, carbon nanofibers (CNFs) were investigated as Pd catalyst support. CNFs were synthesized by the methane catalytic decomposition on a Ni-based catalyst at 700°C. The sulfite-complex method was used to obtain Pd nanoparticles supported on such CNFs and on a benchmark carbon support (Vulcan XC72R, BET 250 m2/g) for comparison purposes. The materials were physico-chemically characterized by using X-ray diffraction (XRD), transmission electron microscopy (TEM), nitrogen physisorption (BET) and thermogravimetric analysis (TGA) in air. The occurrence of highly crystalline graphitic carbon was obtained in the catalysts based on carbon nanofibers as support. TEM images evidenced the filamentous structure of the CNFs with an average diameter around 60 nm and an internal cavity (tubular fibers) of around 15 nm size. The arrangement of graphitic planes corresponded to fishbone-like nanofibers, i.e. showing a certain angle with respect to the growth axis. CNFs presented a BET surface area of ca. 100 m2/g. On the other hand, XRD patterns of Pd catalysts showed the typical fcc structure of palladium, with a crystallite size of around 6.5 nm for both supports (CNFs and Vulcan). The electrochemical behavior of these catalysts for both the oxygen reduction and oxygen evolution reactions was investigated in half-cell configuration in an alkaline solution (6 M KOH). Accelerated degradation tests consisting of potential cycling up to 2 V vs. RHE were also carried out in order to assess the stability of the catalysts under highly corrosive conditions. At the beginning of life, Pd catalyst supported on Vulcan showed slightly higher activity than the Pd/CNF catalyst. The electroactive area obtained by cyclic voltammetry was slightly higher for the Pd/Vulcan. This suggested a more homogeneous distribution of Pd nanoparticles on the surface of commercial support, as a consequence of its higher BET surface area. However, the accelerated degradation test led to a decay of catalytic activity in the case of Vulcan-supported Pd for the ORR, and quite significant for the OER. The decay of activity at the end of life was remarkably lower for the CNF-based Pd catalyst, pointing to a promising stability of the proposed catalyst. Acknowledgements CNR-ITAE authors acknowledge financial support from the European Community’s Seventh Framework Programme (FP7/2010-2013) under grant agreement NECOBAUT n. 314159. M.J.L. thanks the financial support from the Spanish Government throughout the project CTQ2011-28913-C02-01.