Observations of the Earth's climate system over the last century show changes in the atmospheric properties that may be associated with human activities. Growing concerns about these changes, and the desire to predict future states of the climate system as well as its changes led inter alia to the development of the application of general circulation models of the atmosphere and oceans for these types of climate studies. In order to evaluate the correctness and trustworthiness of simulated climate change it is essential to compare modelled changes with observation wherever possible. This is especially important in regions where only little observational data are available, such as the remote Antarctic region.Data of surface and upper air temperatures, longwave radiation, mean sea level pressure, wind speed and sea ice concentration from observational records were compared to the output from a transient experiment of the CSIRO coupled ocean atmosphere sea ice general circulation model. Most of the observational data were taken from the NCEP/NCAR reanalysis project that includes a data assimilation scheme in order to represent irregularly distributed observations onto a regular grid.The examination of changes in the data sets revealed a reasonable representation of observed climate change in the model although there are still many uncertainties in the modelled data as well as in observational records. The models coarse resolution and limited ability to parameterise physical processes result in the lack of confidence in the representation of smaller scale processes. Moreover, the differences between observation and model are often largest for variables that are difficult to observe directly. Many of the uncertainties originate from the shortage of data in the observational record, especially in the Antarctic region. Future improvements of both the observational record (in length and density) and the models (in resolution) are necessary to make confident statements about changes in the climate system.