In the upper stratosphere, an altitude range in which ozone should be in photochemical steady-state, calculated ozone abundances that are derived from a one-dimensional photochemical model with updated chemistry are up to 60% smaller than mean observed values. On the other hand, the model results for the key free radicals (HOx, NOx, and C/Ox species) in the catalytic destruction of ozone are shown to be in reasonable agreement with available measurements. The general validity of the model simulation of C/Ox chemistry is confirmed through a detailed intercomparison between the computed C/O dirunal variation and recently published ground-based microwave observations. Since many field measurements are performed near sunrise or sunset, the uncertainties in the model results arising from the details of the radiation field calculations at large zenith angles are discussed. Although the calculated ozone discrepancy could be the result of a number of errors in adopted photochemical parameters, a sensitivity analysis shows that no reasonable change in any one or two parameters can resolve this problem. The limited available observations regarding the ratio of atomic oxygen to ozone suggest a possible discrepancy in that quantity, which could be responsible for a large part of the ozone problem.