The response of stratospheric O3 to changes in NOx and HOx is tested in a two-dimensional model with the goal of exploring photochemical buffering mechanisms. The model tests are performed under background aerosol conditions at both current (1996) and preindustrial levels of stratospheric halogens. The tests are prompted in part by comparisons to observations from the POLARIS (Polar Ozone Loss in the Arctic Region in Summer) high-altitude aircraft campaign, which show that models significantly underestimate the NOx/NOy ratio in the lower stratosphere but are more successful at reproducing observed O3. Model tests with increased NOx show that increases in NOx-catalyzed O3 loss are subject to photochemical buffering in the lower stratosphere through reduced HOx-catalyzed O3 loss and, at current halogen levels, reduced halogen-catalyzed O3 loss. In contrast, these tests show that O3 is not well buffered with respect to increased NOx in the middle stratosphere, especially at preindustrial halogen levels. Consequently, an NOy increase that reaches the middle stratosphere could lead to modest depletion of column O3, while an NOy increase confined to the lower stratosphere more likely would have a neutral or slightly positive effect on column O3. These results are only mildly sensitive to uncertainties in the NOx/NOy ratio. Additional model tests with increased HOx show some photochemical buffering, but generally lead to mixed effects on O3 which depend on the nature of the HOx source gas. This source gas dependence is especially strong at current halogen levels. ¿ 1999 American Geophysical Union