Lithium-oxygen has one of the highest specific energies among known electrochemical couples and holds the promise of substantially boosting the specific energy of portable batteries. Mechanistic information of the oxygen reduction reaction by Li (Li-ORR) is scarce, and the factors limiting the discharge and charge efficiencies of the Li-oxygen cathode are not understood. To shed light on the fundamental surface chemistry of Li-ORR, we have performed periodic density functional theory calculations in conjunction with thermodynamic modeling for two metal surfaces, Au(111) and Pt(111). On clean Au(111) initial O2 reduction via superoxide (LiO2) formation has a low reversible potential of 1.51 V. On clean Pt(111), the dissociative adsorption of O2 is facile and the reduction of atomic O has a reversible potential of 1.97 V, whereas the associative channel involving LiO2 is limited by product stability versus O to 2.04 V. On both surfaces O2 lithiation significantly weakens the O–O bond, so the product selectivity of the Li-ORR is monoxide (LixO), not peroxide (LixO2). Furthermore, on both surfaces LixO species are energetically driven to form (LixO)n aggregates, and the interface between (LixO)n and the metal surfaces are active sites for forming and dissociating LiO2. Given that bulk Li2O(s) is found to be globally the most stable phase up to 2.59 V, the presence of available metal sites may allow the cathode to access the bulk Li2O phase across a wide range of potentials. During cycling, the discharge process (oxygen reduction) is expected to begin with the reduction of chemisorbed atomic O instead of gas-phase O2. On Au(111) this occurs at 2.42 V, whereas the greater stability of O on Pt(111) limits the reversible potential to 1.97 V. Therefore, the intrinsic reactivity of Pt(111) renders it less effective for Li-ORR than Au(111).