In this work, a new methodology is applied to study hot nights, also called tropical nights, in the metropolitan area of Madrid. To evaluate hot nights from a temporal and spatial perspective in which the population may be affected by thermal stress, high resolution gridded hourly temperature data were used. The use of two indicators obtained through hourly data, together with the climatic information provided by the UrbClim model, allowed to evaluate the thermal night characteristics of July between 2008 and 2017 at a detailed scale. Hence we were able to estimate precisely the risk to the well-being and health of the population. The results show great interurban variability in terms of intensity and duration of heat stress and a significant correlation between heat island intensities and excess heat. Likewise, a close relationship between the typologies of land uses and urban structures defined in the Urban Atlas and the indices of night heat excess has been established.