Chlorine activation and ozone depletion in the Arctic winter stratosphere of 1996--1997 have been studied with a newly developed stratospheric chemistry-transport model (CTM). The chemistry scheme, using a Euler Backward Iterative approximation method, includes a comprehensive set of reactions on ternary aerosol and ice particles, which has been tested against a numerically exact solver. Tracer transports in the CTM are calculated from European Center for Medium-Range Weather Forecasts (ECMWF) meteorological analyses. Comparisons have been made with O3 and ClO measurements, and with ozone loss rates derived from observations during February and March 1997. ClO production and ozone depletion are somewhat underestimated by the model. Furthermore, uncertainties regarding the aerosol phase are tested. Assuming nitric acid trihydrate (NAT) particles to form at their melting point, while liquid aerosol is present simultaneously in the model, gives rise to the largest ClO production and the strongest ozone depletion. By correcting for an ECMWF temperature warm bias we obtain a similar large effect on calculated ClO production and ozone depletion for the 1996/1997 Arctic winter, whereas uncertainties in the chlorine abundance seem less important. An average warm bias of 1.3 K at polar stratospheric cloud temperatures on the 50 hPa model level reduces the calculated ozone depletion rates over February and March by 35%. Observations of ClO are reproduced when lower temperatures and maximum Cly abundance are assumed, but ozone depletion is slightly overestimated in that case. ¿ 2000 American Geophysical Union