To design effective strategies for sustainable urban resource management, it is essential to understand which urban characteristics underlie consumption patterns. We used multiple linear regression analyses to examine sixteen factors on their explanatory power for spatial variation in residential electricity, gas and water consumption in Amsterdam. Four models per resource were used, based on distinct spatial units aggregating high-resolution data: neighbourhoods, districts, 100 m squares and 500 m squares. We found twelve explanatory variables for spatial variability in consumption in total and nine or ten per resource. The number and relative importance of explanatory variables varies with the spatial units used. Overall, neighbourhood models explain variance in consumption data best (adjusted R² = 0.88, 0.86, 0.74). Income level and building type stand out for having high relative importance (top 4) in all four models for two of the three resources; migration history shows an important correlation with water consumption, which was not described hitherto. We conclude that explanatory variables for resource consumption are sensitive to size and shape of spatial units used. We recommend to use future high resolution studies for different resources of interest to determine which spatial and temporal resolutions of analysis can support urban planners and designers in formulating context-specific interventions.