A unique approach has been applied to evaluate the performance of Antarctic Mesoscale Prediction (AMPS) forecasts for the summer season. The AMPS forecasts were compared to automatic weather station (AWS) observations, and model validation statistics have been calculated from this comparison. Model validation statistics have been calculated for both seasonal time periods (NDJ) and also for different synoptic weather patterns identified by an artificial neural network technique known as self-organizing maps (SOMs). Results from this study are expected to be of use to both operational weather forecasters and to model developers.