An off-line three-dimensional stratospheric chemical transport model (CTM) has been developed and integrated in a series of 6-year experiments covering 1991 to 1997. The model has a detailed chemistry scheme and is forced by the meteorological analyses for the temperatures and horizontal winds, while the vertical motion is diagnosed using a radiation scheme. By using the meteorological analyses, the model captures the interannual variability in chlorine activation in the Arctic winter lower stratosphere, as well as the much more regular, strong activation in the Antarctic. The model fields show generally good agreement with long time series of ground-based observations. In particular, the model simulations of column O3 are excellent in terms of capturing the magnitude and day-to-day, seasonal, and interannual variations at all latitudes. However, a large model/observation discrepancy for O3 occurs at high latitudes during the summer, when the model overestimates the O3 profile throughout the lower and mid-stratosphere. The model also reproduces many features in ground-based observations of HCl, ClONO2, HNO3, and NO2, such as short-term variability, the seasonal cycles, and winter/spring enhancements of ClONO2 at northern midlatitudes. By running the model without cold, chlorine activating heterogeneous reactions the effect of polar (and subpolar) processing on midlatitudes has been estimated. Polar processing results in 2--3% less O3 at 50 ¿N throughout the year, and around 5% less O3 at 50 ¿S. Direct chlorine activation on enhanced midlatitude aerosol contributes 1% to column ozone depletion at mid and low latitudes in early 1992. The enhanced aerosol also caused column ozone reductions of around 3% at 45 ¿N throughout 1992 and 1993. However, the CTM runs confirm the suggestion interannual dynamical variability (which may be partly driven by the radiative effects of the aerosols) contributed to the large negative ozone anomaly in northern midlatitudes in 1992/1993. ¿ 1999 American Geophysical Union