We have developed a global three-dimensional transport-chemical model of the stratosphere (called the Study of Transport and chemical Reactions in the Stratosphere (STARS)) which includes a representation of the formation of polar stratospheric clouds (PSCs) and heterogeneous reactions on the surfaces of PSCs and sulfate aerosols. The formation of the observed springtime Antarctic ozone hole is well reproduced by the model. A maximum of 40% total ozone depletion occurs in October. Calculated ozone and chlorine concentrations are consistent with satellite observations. After the breakdown of the polar vortex in December, air with depleted ozone is transported to midlatitudes in the southern hemisphere, resulting in a 2--4% ozone decrease at 50 ¿S in December and a 1% decrease in the subtropics. Ozone-poor air masses are also transported to the troposphere and produce a significant decrease in upper tropospheric ozone. Model calculations show that the calculated ozone depletion is not significantly modified when type I PSC particles are assumed to be liquid ternary solutions (H2SO4/HNO3/H2O) rather than solid nitric acid trihydrates. Ice particles (type II PSCs) sediment into the troposphere, producing a large decrease in the concentrations of stratospheric HNO3 and NO2. As a result, the conversion of ClO into ClONO2 is reduced, and the concentration of reactive chlorine remains high for two additional weeks. The ozone column abundance is reduced by an additional 10% during the month of October. The model results show that the ozone minimum observed in Antarctica several decades ago (preindustrial chlorine levels) is produced by (natural) dynamical processes. Under these conditions the polar ozone depletion caused by chemical processes was very small (maximum of 3%) in October. In November the ozone concentration even increased above 22 km in response to PSC processes.¿ 1997 American Geophysical Union