Increasingly, many aspects of the study of Antarctica and the high southern latitudes are being aided by various types of numerical models. Among these are the General Circulation Models (GCMs), which are powerful tools that can be used to understand the maintenance of present atmospheric climate and determine its sensitivity to imposed changes. The changes in the ability of GCMs used over the last two decades to simulate aspects of atmospheric climate at high southern latitudes are traced and it is concluded there has been a steady improvement in model products. The task of assessing model climates in high southern latitudes is made difficult by the uncertainties in the data used for the climatological statistics. It is suggested that the quality of the climates produced by most modern GCMs in many aspects cannot be said to be poor, especially considering the uncertainties in ‘observed’ climate. There is obviously need for improvements in both modelling and observations. Finally, some topics are highlighted in which the formulation of models could be improved, with special reference to better treatment of physical processes at high southern latitudes.