A three-dimensional chemical transport model is utilized to study the transport out of the Antarctic polar vortex during the southern hemisphere spring. On average, over five consecutive years between 1993 and 1997, horizontal transport out of the vortex into the midlatitude stratosphere is smaller than vertical transport into the troposphere. However, there is significant interannual variability in the magnitude of mass exchange, which is related to year-to-year fluctuations in planetary wave activity. In 1994 the net loss of the vortex tracer mass in September is similar to that in October. However, the relative mass flux entering the midlatitude stratosphere and the troposphere differ between the two months. The ratio of horizontal transport out of the vortex to vertical transport into the troposphere is about 3:7 in September and 5:5 in October, indicating the higher permeability of the vortex in October compared to September. The September mass flux into the troposphere is larger than in October, consistent with the fact that stronger diabatic cooling occurs in September than October over Antarctica. The estimated ozone change at southern midlatitudes due to the intrusion of ozone-depleted air from high latitudes during September--October 1994 is about -0.44% per decade, which could contribute up to 10% of observed ozone decline at southern midlatitudes in spring. This amount is an underestimate of the dilution effect from high latitudes during the spring season, as it does not include the vortex breakup in late spring.