Aspects of the zonally averaged simulation produced by the middle atmosphere version of the latest generation of the National Center for Atmospheric Research (NCAR) Community Climate Model (CCM2) are discussed together with its dependence on horizontal resolution. This model is serving as the basis for several studies of stratospheric transport, chemistry, and dynamics, so the quality of its basic simulation is of considerable interest. It is shown that the model produces a reasonable simulation of the annual cycle of the northern hemisphere and significantly improves upon previous versions of the model. Serious deficiencies exist in the simulation of southern hemisphere winter stratosphere. The polar vortex is much too strong in southern winter, with winds speeds almost double those observed. The strength of the vortex decreases modestly with increasing horizontal resolution, but the vortex is still much too strong at the highest resolution (T106) which can currently be used even for short simulations. This problem is common to other middle atmosphere general circulation models and CCM2 does not represent a significant advance in this respect. The forcing of the mean flow by the Eliassen-Palm (EP) flux divergence of the resolved waves is much larger during winter in the northern hemisphere stratosphere than in the southern hemisphere. Nevertheless, the interhemispheric difference in the strength of the simulated stratospheric polar vortex requires the effect of the parameterization of stationary gravity waves generated by flow over orography. When this parameterization is removed, the northern vortex in January reaches comparable strength to that of the southern vortex in July. Based on the current simulations and comparisons with parameterizations in two-dimensional models, it is argued that a parameterization including the effects of nonstationary gravity waves from sources other than flow over orography will be required to obtain accurate simulations of the Antarctic vortex. The alternative of direct simulation of the gravity waves with high-resolution models will only be feasible for test simulations in the near future. ¿ American Geophysical Union 1995