Fuel cells are important energy conversion devices that provide an alternative energy source for transport, stationary and portable applications. Electricity is generated by harnessing coupled oxidation-reduction reactions of various chemistries, in each case limited by the sluggish oxygen reduction reaction (ORR) performed at the fuel cell cathode. Currently, platinum on carbon (Pt/C) is the standard material applied as the catalyst for ORR. However, Pt is expensive and scarce, drawing much interest in developing non-Pt catalysts. Carbon nanomaterials, including fullerenes, carbon nanotubes (CNTs), graphene, and their composites, have been actively studied as a potential candidate to replace Pt/C. Carbon-based ORR catalysts have been demonstrated to be competitive with Pt/C in electrochemical efficiency and durability, particularly in alkaline fuel cells (AFC). These sp 2-bonded carbon nanostructures are inherently constituted by extended p-conjugation systems, making them essentially unreactive toward molecular oxygen (O2) in their pristine forms. However, the conjugated p-systems can be significantly disrupted with localized p electrons through dopant incorporation, defect creation, and/or local lattice distortions. The doped, defective and distorted carbon structures are currently thought to be responsible for O2 adsorption/activation and thus ORR, but the active sites and the exact mechanisms are still uncertain. Using density functional theory (DFT) calculations, we have identified a novel method for the activation of graphene-like structures. This modification creates quasi-localized electron states in the π-conjugated system which are vulnerable to oxidation via O2. The result is favorable O2 adsorption, which despite being necessary for ORR activity is not predicted for previously proposed active sites. We have identified the importance of using proper solvation models at the solid-liquid interface. While many previous DFT studies were modeled in the gas phase where O2 adsorption is consistently predicted to be endothermic on carbon, our study clearly demonstrates that a more accurate description of interactions at the carbon-solvent interface is necessary to properly describe O2 adsorption. In addition we have found that more detailed descriptions of charge injection are necessary at carbon-based catalysts, where polarization of the carbon electrode is significant. Ab-initio MD simulations were performed with an explicit solvent model to outline a full reaction mechanism identifying elementary reaction steps. The rate limiting step of the ORR in this system is predicted to be desorption of the final reduction products for which we predict a moderate but surmountable activation barrier. Excitingly, we predicted overpotentials for the elementary reaction step of the modified carbon structures within the range of 0.4-1.1 eV, indicating our model can be used to design materials to compete with Pt/C. Using our model of the carbon-solvent interface, we revisited the prevalent types of activated carbon, including doping and plane edge structures, providing an improved description of the ORR activity. Our study provides insight into molecular mechanisms underlying the superior ORR activity of nanostructured carbon materials and also guidelines for the rational design and development of high-efficiency, cost-effective carbon-based ORR catalysts for fuel cell applications.