The management of large building stocks with limited resources poses a problem of prioritization of refurbishment actions. Also, available information is often incomplete and the process of standardization and updating is expensive and time consuming. Some public administrations are developing BIM models of their stock, but they want to limit the complexity of the models within the lowest amount of information needed for management and maintenance purposes, in order to make that process affordable. However, they cannot miss to comply with the regulations in force and to plan regular maintenance, which requires a reliable knowledge of facilities. For that reason, this paper presents a decision support tool that can help prioritize refurbishment actions on large building assets. To this purpose, many requirements must be jointly considered in this examination, hence a multi-criteria analysis is an efficient methodology to find out the best scenario. In order to deal with the extensive and uncertain information that must be managed in this process, the decision support tool is built on Bayesian Networks. In addition, the minimum level of detail that must be provided by the BIM models of the stock is discussed, so that the information needed as input to the decision tool can be read automatically from BIM models. Finally, the several indicators corresponding to different requirements are compared by means of a multi-criteria analysis tool, that is able to weigh different criteria. As a result, each building is assigned an overall score and a ranking is created. The lowest the score assigned to the building the most urgent refurbishment actions are. In this paper, the application of the decision tool on a real test case will be presented.