The presence of appreciable molecular oxygen at the location of maximum temperature has been observed in non-premixed oxygen-enhanced combustion (OEC) processes, specifically in flames having a high stoichiometric mixture fraction (Z(st)) produced with diluted fuel and oxygen-enrichment. For conventional fuel-air flames, key features of the flame are consistent with the flame sheet approximation (FSA). In particular, the depletion of O-2 at the location of maximum temperature predicted by the FSA correlates well with the near-zero O-2 concentration measured at this location for conventional fuel-air flames. In contradistinction. computational analysis with detailed kinetics demonstrates that for OEC flames at high Z(st): (1) there is an appreciable concentration of O-2 at the location of maximum temperature and (2) the maximum temperature is not coincident with the location of global stoichiometry, O-2 depletion, or maximum heat release. We investigate these phenomena computationally in three non-premixed ethylene flames at low, moderate, and high Z(st), but with equivalent adiabatic flame temperatures. Results demonstrate that the location Of O-2 depletion occurs in the vicinity of global stoichiometry for flames of any Z(st) and that the presence of appreciable O-2 at the location of maximum temperature for high Z(st) flames is caused by a shift in the location of maximum temperature relative to the location Of O-2 depletion. This shifting is attributed to: (1) finite-rate multi-step chemistry resulting in exothermic heat release that is displaced from the location of O-2 depletion and (2) the relative location of the heat release region with respect to the fuel and oxidizer boundaries in mixture fraction space. A method of superposition involving a variation of the flame sheet approximation with two heat sources is shown to be sufficient in explaining this phenomenon. (C) 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved.