Abstract: The planning, construction, management and use of our built environment are affected by diverse social, economic and environmental factors. Sustainable urban development is dependent on the understanding of the complex relations between the built environment, the social activities that take place over time and the interaction with the natural environment. The challenge to understand urban systems on both the local and global scale has inspired researchers and national agencies to develop sustainability indicators to support the planning, construction, management and use of the built environment. Access to open data of our built environment in national, regional and local databases opens new possibilities to generate models of our urban systems to facilitate visualization and analysis of indicators in order to enhance awareness of sustainability dimensions. Here spatial Extract, Transform and Load (ETL) technologies can be used in combination with Geographic Information system GIS to manage data sets from multiple sources in different formats. The purpose of this research is to investigate how spatial ETL technologies can be used to develop models in order to analyse and visualize the performance of urban systems. The applied method is grounded in system development and based on an abductive research approach that was repeated in six studies. Three of the studies deal with the relocation of Kiruna where models of the city was created and used to investigate the impact of mining subsidence on energy supply, infrastructure and buildings. The fourth case investigates the selection of insulation material on the embedded energy in a passive house in Kiruna. In the fifth case an urban model of the twin towns Malmberget/Gällivare was created to explore and relate data on attitudes from a survey to public data on population, infrastructure and built environment. The final case is the development of an energy atlas containing 90% of the multifamily building stock in Sweden. The atlas combines the energy performance and renovation status of multifamily buildings with public data of ownership, income of residents etc. for individual buildings in 3D models or aggregated on spatial scales ranging from 250x250 m squares through district and municipality to county areas in Sweden. The result shows that multiple sources in different formats, both standardized and non-standardized, can be utilized in the extraction of information for the purpose of developing urban performance models. The Swedish high-resolution LiDAR digital height model together property information makes it possible to represent the built environment by extruded footprints to give a 3D representation of all urban areas in Sweden (Level-Of-Detail 1). In combination with performance data (e.g. energy use, renovation status or result from surveys) urban performance GIS models can be created and visualized in applications (such as Google Earth, 3D pdf) to support decision-making on both individual and institutional level. The automation of the process to develop performance models offers a method for customizing information deliveries on the fly using original data sources according to defined requirements. The flexibility and customization are kept in the process rather than in the delivered model. This makes it easier to keep the performance model up to date. For the management of large performance models, e.g. the example of the national energy atlas, a staging phase was added in the automation process, in order to reduce the processing time.