A three-dimensional spectral chemical transport model truncated at T21 is employed to simulate N2O transport. The wind and vertical motion fields are taken from the U.K. Meteorological Office four-dimensional assimilation data set. UARS cryogenic limb array etalon spectrometer (CLAES) N2O measurements are used to initialize the model in late August 1992. Model results are shown to simulate the CLAES measurements quite well over the first few months: N2O variability is similar at extratropical latitudes in the southern hemisphere over the period September 2--17, 1992, at 4.6 and 10 mbar, and there is good agreement in the synoptic maps of minor warmings during this period. Prior to a large warming event on September 30, minor stratospheric warmings are shown to produce negligible changes in the vortex below 4.6 mbar, but considerable mixing of air from the vortex edge and subtropical air is indicated. This results in a steepening of the N2O gradient at the vortex edge. During a warming event when the vortex center moves away from the pole, downward transport by the residual circulation can be large. This is offset by eddy transport effects, but these terms reverse during the recovery from the warming. From September 2 to 17, there is evidence of continuous mixing at midlatitudes at 4.6 mbar in contrast to more discontinuous, warming-associated mixing at 10 mbar. The breakup of the vortex is initiated by the September 30 warming, and a warming on October 13 has a strong influence on the breakup. The breakup propagates downward. The climatological distribution of N2O in the tropics follows the seasonal variation of the solar radiation with a maximum, which is determined by the strength of the upward residual motion, shifting towards the summer hemisphere by 10¿--15¿ latitude. The surf zone in both the model and the observations at the middle latitudes is well defined, but the gradients of N2O at the edge of the tropics and at the edge of the vortex are smaller in the model than in the observations. This is probably being caused by excessive mixing in the model.¿ 1997 American Geophysical Union