To investigate the role of ice dynamics on the seasonal cycle of Antarctic sea ice, a series of 2-year simulations of the Weddell Sea pack ice are carried out. The simulations employ the dynamic thermodynamic model developed by Hibler (1979) and use a 1-day time step on an 18¿15 grid with a resolution of 222 km. Daily atmospheric data from 1979 are used to drive the simulations. With atmospheric temperature fields modified to reflect buoy observations, the simulations yield a seasonal cycle of ice with maximum and minimum extents close to that observed. Except for portions of the western Weddell, the advance of the ice is found to be primarily thermodynamic in nature, while the rapid decay depends critically on the presence of both leads and lateral ice advection. The average fraction of open water is substantial even in winter, and varies from a minimum of 10% in September to a maximum of 35% in March. These values are in general agreement with estimates from satellite microwave data. In early summer the ice motion causes a residual tongue of ice to extend eastward from the Antarctic Peninsula in agreement with observations. Mean ice thicknesses are consistent with observations and vary from aboaut 3 m in the perennial ice in the western Weddell to 1 m in first-year ice in the eastern Weddell. Possible feedback between the ice extent and atmosphere are investigated by simulations with the surface air temperatures constrained to below-freezing values in the presence of ice. These 'coupled' simulations yield a more realistic rapid decay in summer with dynamics included, but an unrealistically small decay without dynamics. Simulated ice drift results yield mean drift rates of about 5 km/day, in good agreement with buoy drift observations but with slightly inadequate northward transport in the western Weddell. Near the ice edge the drift rates are relataively insensitive to the ice strength. Near the coast, however, lower strengths are found to yield a decrease in northward drift rates.