At present, all over the world, the assessment of the performance and safety of dams, their diagnosis and proposals for corrective actions are carried out by expert engineers during dam reviews. Performance assessment is a difficult task, in particular for dams: the loss of dam performance is the result of a succession of phenomena, stemming from miscellaneous and complex sources that lead to as many miscellaneous and complex consequences, ranging from the deterioration of one or more functions to complete dam failure. To perform dam reviews, experts handle multiple heterogeneous data: visual inspections, data from monitoring devices, outputs of mathematical models, and design and construction data. These data usually present imperfection features: imprecision, uncertainty, incompleteness. Here, we propose using two approaches capable of modelling, aggregating heterogeneous imperfect information and representing expert knowledge to develop tools and methods for managing dam performance. The two approaches are Knowledge-Based Systems and possibility theory. This decision-support system is developed according to two steps: (i) the formalisation and aggregation of heterogeneous information from different sources in adequate models, (ii) the representation and propagation of imperfections into these models and the expression of outputs in suitable formats to facilitate decision-making and improve the communication of results. The method was applied to embankment and gravity dams and their foundations.