Climate change significantly affects the operating environment of buildings. These changes impact both energy efficiency and occupants’ comfort and remain crucial even in building restoration, where design decisions typically rely on historical data, yet performance depends on anticipated future scenarios. The present work evaluates the impact of different climate datasets on dynamic energy simulations for an educational building in Central Italy, focusing on estimating heating demands across historical, current, and future climatic scenarios. The assessment considers both the building’s current state and potential energy-efficient retrofits. Initially, various meteorological datasets, including measured and model-generated data, are selected to predict key weather parameters. The analysis reveals the potential and limitations of regional climate models (RCMs) in estimating these variables, with the MM5 dataset emerging as the most reliable. Subsequently, the energy performance of the reference building and its vulnerability to climate change are assessed. Our results show significant differences in energy demand based on construction periods, with the oldest section consuming 29% to 54% more energy monthly than the newer sections. Moreover, using non-representative climatic files can lead to prediction errors of up to 199%. Finally, the building’s energy behaviour is analysed under future climate conditions by generating typical meteorological years (TMYs) for 2030, 2050, and 2070. This analysis evaluates the energy requirements for both existing and retrofitted building configurations. The findings confirm that retrofit interventions with high-performance insulation and upgraded windows significantly enhance the building’s energy efficiency and resilience to future climate conditions, leading to annual energy savings of 50% to 57%.