The characteristics of urbanism and its construction properties are critical factors in shaping energy demand and citizens' quality of life. Urban typology approaches classify the urban built environment through common patterns, facilitating building energy analysis and energy retrofitting. However, current methods struggle between detailed local building classifications and broad land spatial distributions, which include built environments. This paper presents a systematic methodology to objectively identify and classify residential urban areas according to representative Homogeneous Urban Zones (HUZs), aiming at improving the analysis of urban block energy behaviour by considering buildings and their surrounding urban conditions. Based on the characterisation of urban clusters, this methodology merges current urban classifications and defines representative urban block indicators using geographic information systems. The urban clusters are statistically delimited by testing a set of urban indicators grouped by constructive systems, morphology, and surface cover. If they successfully meet the statistical criteria of the outlier's filter, test-of-goodness-fit, and non-overlapping check, the clusters are then validated as HUZs; otherwise, they are merged, split, or rejected. Finally, the urban residential areas classified as HUZs are characterised by average, maximum, and minimum range values based on a set of representative urban indicators. This methodology was applied to the city of Madrid resulting into the definition of 10 statistically significant HUZs, grouped into four different categories. Because of the shown good adaptability of HUZs to the local urban context, it is expected that the present HUZ method represents an improvement for the standardisation of the residential building energy modelling evaluation.