The chlorine activation and subsequent ozone loss of the northern winter lower stratosphere have been modelled using different schemes for type I polar stratospheric clouds (PSCs) and sulphate aerosols. Type I PSCs were assumed to consist of either nitric acid trihydrate (NAT) at equilibrium, supercooled ternary solutions (STS) at equilibrium, or to follow a hysteresis cycle between frozen and liquid particles depending on the temperature history. The sulphate aerosol was assumed to be present as either liquid binary H2SO4/H2O aerosol (LBA) or as solid sulphuric acid tetrahydrate (SAT). Our box model integrations show that NAT and STS, representing the upper and lower limits of lower stratospheric chlorine activation, respectively, appear to destroy ozone equally efficiently after a cold PSC event (Tmin≤190 K at 50 mbar). For higher minimum temperatures, up to the equilibrium NAT point, there is significantly more ozone loss in the NAT scheme than in the STS scheme. On NAT, chlorine is activated directly by ClONO2+HCl→2Cl+HNO3, whereas on STS, indirect activation by ClONO2+H2O→HOCl+HNO3 followed by HOCl+HCl→2Cl+H2O, dominates. During the processing period, the indirect activation on STS will produce a temporary peak in HOCl. Box model integrations also show that direct chlorine activation is faster on SAT than on LBA, yielding significantly more ozone loss in air parcels which remain below the SAT melting point (215--220 K). Our single-layer chemical transport model simulations (&thgr;=465 K) of the lower stratospheric chlorine activation during Arctic winter 1994/1995 show that chlorine is activated more quickly on NAT than on STS. However, in mid December 1994, when temperatures are low enough for substantial STS particle growth, maximum active chlorine becomes similar in both schemes and remains similar until the end of January 1995. A model integration which includes SAT produces up to 200 parts per trillion by volume more ClOx, inside the polar vortex during Arctic winter 1994/1995, than a model integration which includes LBA. The high melting point of SAT means that it may contribute to midlatitude ozone loss when filaments of processed air are shed by the vortex. For example, a model integration shows that air peeling off the Arctic vortex on February 14, 1995, contains 10% more ClOx at middle latitudes in an integration that includes SAT formation in a hysteresis scheme, than in an integration that includes LBA. The major differences in ozone loss predicted by the model PSC schemes occur inside the polar vortex. The largest differences in ozone at 465 K inside the vortex at the end of March 1995, up to 500 parts per billion by volume, are found between the equilibrium NAT and STS schemes. Ozone values in the hysteresis schemes are intermediate between those of the NAT and STS schemes. The inclusion of SAT in a hysteresis scheme does not have a major global effect. ¿ American Geophysical Union 1996