This study models maximum temperatures in Switzerland monitored in twelve locations using the Generalised Extreme Value (GEV) distribution. The parameters of the GEV distribution are determined within a Bayesian framework. We find that the parameters of the underlying distribution underwent a substantial change in the beginning of the 1980s. This change is characterised by an increase both in the level and the variability. We assess the likelihood of a heat wave of the Summer of 2003 using the fitted GEV distribution by accounting for the presence of a structural break. The estimation results do suggest that the heat wave of 2003 appears not that statistically improbable event as it is generally accepted in the relevant literature.