Abstract Given that the existing residential buildings are expected to become a huge part of the future stock due to their low replacement rate, retrofitting plans are crucial to meet 2030 and 2050 energy efficiency targets. Notwithstanding, an extensive assessment of the current energy and thermal performance of the stock must be conducted prior to the proposal of energy saving measures in order to properly tackle the retrofit process. Thus, the analysis and characterisation of the existing buildings under real variability conditions through statistical techniques is key to provide useful information at the stock level, instead of the most commonly single-building level approach. In the presented study, a statistical analysis on the most predominant variability ranges of the social housing stock of southern Spain (Andalusia) is carried out. Efforts are focused on the building characterisation of the linear block typology. To do so, an extensive database, which contains information on slightly under 39,500 social dwellings, is analysed. The conclusions reported in this study may be implemented into the construction of real case building archetypes through bottom-up building stock modelling techniques with the objective of assessing the real energy and thermal performance of the existing stock, providing useful information for public stakeholders.