Abstract This paper presents a Bayesian approach for developing city-scale energy models of the built environment and demonstrates its application to non-domestic buildings in Greater London. The work draws upon available information of the building stock, such as: mapping databases, floorspace statistics, energy benchmarks, and measured energy consumption reported in display energy certificates of public buildings. The resulting model is able to describe the spread due to variation of energy consumption across buildings within a similar category. These spreads (or distributions) can be used for estimating the probability distribution of the gross energy consumption per local authority in Greater London. The work is driven by the need to quantify future energy demand of buildings in their urban context as a function of projected growth of buildings and populations, refurbishments, policies incentivizing energy efficiency measures, and changes in building operation. The focus on the non-domestic sector enables a framework that accommodates diverse set of activities and uses of buildings within an urban region.