A three-dimensional mechanistic model of the middle atmosphere with calculated dynamics and chemistry (Rose and Brasseur, 1989) is used to study the behavior of chemically active trace gases at high latitudes in winter and spring, and to simulate the formation of an ozone hole in Antarctica. The dynamics of both hemispheres is simulated by applying at the lower boundary of the model (8.5 km) a wavelike perturbation representing qualitatively a climatological tropospheric forcing. The chemical heterogeneous processes converting chlorine reservoirs into active chlorine in cold air masses are parameterized. The model simulates the behavior of nitrogen oxides, nitric acid, water vapor, methane, hydrogen radicals, chlorine compounds, and ozone. It reproduces important features observed during different Antarctic and Arctic observation campaigns. The ozone hole in the southern hemisphere can only be simulated when the heterogeneous polar chemistry is taken into account. The springtime ozone depletion over Antarctica calculated in the model is thus mostly the result of chemical removal although the dynamics is responsible for the low temperature that triggers the large ozone loss rates. Unresolved questions are related to the strength of the vertical exchanges inside the vortex, the preconditioning of trace gases before and during the winter season, the behavior of the different trace gases as the vortex breaks down (dilution effects), accurate determination of the ozone sink inside the vortex, and a better quantitative estimation of the role of polar stratospheric clouds. Despite elevated concentrations of active chlorine at high latitudes in the northern hemisphere in late winter, no ozone hole is produced by the model, even with chlorine levels as high as 6 ppbv. This conclusion could, however, be modified for very stable and cold winters with delayed final warming. ¿1991 American Geophysical Union