Climate model data provide large, dense coverage and long time-series, characteristics that are advantageous in the study of climate. However, it is not recommended that such data be used in any region without prior evaluation of their reliability based on comparisons with in situ observations. In this study, the accuracy of maximum and minimum temperature data from the ALADIN-Climate regional climate model (with a 25-km horizontal resolution driven at the lateral boundaries by the ERA-40 reanalysis) have been assessed for 53 stations across the Balkan Peninsula. The model temperatures corresponding to each station were extracted from their nearest land grids. The model data were first compared with observations and subsequently examined for their ability to identify extreme temperature events. In general, the model was found to be quite accurate in describing the seasonal cycle, as well as simulating the spatial distribution of temperature. Simulations were more realistic for stations along coastlines, highlighting the constraints of the topographic forcing in the simulations. Assessing the performance of the model to determine extremes (warm and cold spells), it was found to be better at detecting cold spells and has a tendency toward overestimating the frequency of occurrence of warm spells, particularly in summer.