An observationally based, mechanistic dynamical model is used to simulate the decline of total ozone during September and October for the years 1979 through 1986. Vertical velocities derived from observed stratospheric temperature changes and computed radiative heating rates are used to advect an ozone mixing ratio profile during the Antarctic spring period. An early August 1982 Syowa balloonsonde ozone profile is used to initialize the computations. The model reasonably simulates the September and October changes in total ozone, considering the uncertainties in the observed data and the radiative heating. The simulated decline is found to be very sensitive to the choice of initial ozone profile and to small changes in the radiative heating. The results of this study suggest that the dynamical hypothesis of the Antarctic ozone depletion is both quantitatively credible and consistent with the observed temperature changes. ¿ American Geophysical Union 1988