Changes in weather patterns worldwide and global warming increased the demand for high-performance buildings resilient to climate change. Building Performance Simulation (BPS) is a robust technique to test, assess, and enhance energy efficiency measures and comply with stringent energy codes of buildings. Climate has a considerable impact on the buildings' thermal environment and energy performance; therefore, choosing reliable and accurate weather data is crucial for building performance evaluation and reducing the performance gap. Typical Weather Years (TWYs) have been traditionally used for energy simulation of buildings. Even if detailed energy assessments can be performed using available multi-year weather data, most simulations are carried out using a typical single year. As a result, this fictitious year must accurately estimate the typical multi-year conditions. TWYs are widely used because they accelerate the modeling process and cut down on computation time while generating relatively accurate long-term predictions of building energy performance. However, there is no certainty that a single year can describe the changing climate and year-by-year variations in weather patterns. Nowadays, with increased computational power and higher speeds in calculation processes, it is possible to adopt multi-year weather datasets to fully assess long-term building energy performance and avoid errors and inaccuracies during the preliminary selection procedures. This study aims to investigate the impact of Typical Weather Years and Actual Weather Years (AWYs) on a single-family house and a university building under two opposite climates, Winnipeg (cold) and Catania (hot). First, a single-family house in Winnipeg, Canada, was selected to evaluate how typical weather years affect the energy performance of the building and compare it with AWYs simulation. Two widely used typical weather data, CWEC and TMY, were selected for the simulation. The results were compared with the outcomes of simulation using AWYs derived from the same weather station from 2015 to 2019, which covered the latest climate changes. The results showed that typical weather years could not sufficiently capture the year-by-year variation in weather patterns. The typical weather years overestimated the cooling load while underestimating the heating demands compared to the last five actual weather years. A more extensive study was conducted for more confidence in the findings and understanding of the weather files. The research was expanded by comparing the results of building performance simulation of the single-family house and an institutional building with more complex envelope characteristics belonging to the University of Manitoba under cold (Winnipeg, Canada) and hot (Catania, Italy) climates. Overall, 48 simulations were performed using ten actual weather years from 2010 to 2019 and two TWYs from each climate for both buildings. The results showed that while the TWYs either overestimate or underestimate the cooling and heating demands of both buildings, cooling load predictions were highly overestimated in the heating-dominant climate of Winnipeg, ranging from 10.5% to 82.4% for both buildings by CWEC and TMY weather data. In the cooling-dominant climate of Catania, energy simulations using IWEC and TMY typical weather data highly overestimated the heating loads between 2.8% and 82.4%.