The health risks posed by hot weather are growing as increasingly frequent extreme weather is brought about by climate change. People spend upwards of 80% of time indoors and so human health is largely dependent on the internal environment of buildings. In the building industry engineers currently design buildings for high-energy performance by maximising heat retention, and whilst this may be effective in cold winters, it can lead to unbearable indoor conditions in hot summers. Thermal comfort inside buildings is a well-discussed topic both in industry and academia, but absolute peak thresholds, especially for heat stress still require development. In this thesis the outcomes of research into the effects of current and future hot weather on the heat stress of occupants inside buildings are presented. Hot weather data from the current climate and mortality rates are compared and several temperature metrics are analysed with respect to health risk forecasting performance, so that peak threshold limits for human health indoors are established for the building design industry. Reference weather data used in building simulations for health assessment is currently chosen based on air temperature alone. In this thesis new reference weather data is created for near-extreme and extreme weather and for current and future climates, based on the peak threshold metric research and future weather analysis. By 2050 hot weather reference years currently occurring once every seven years could become an annual occurrence, and by 2080 extreme hot weather reference years currently occurring once in twenty-one years could become an annual occurrence. Computational fluid dynamics is then used to simulate the internal heat stress inside a building model, and a surrogate model is created to emulate heat stress levels for full calendar years of future climates for several UK locations. It is envisaged that the results presented in this thesis will help inform the industry development of new reference data and aid better building design.