For fire safety studies in nuclear facilities, IRSN uses the SYLVIA software to simulate fire scenarios in highly confined and mechanically ventilated compartments and airborne contamination transfers inside nuclear facilities. In order to take into account the different sources of uncertainty resulting from initial and boundary conditions as well as from model parameters, the SYLVIA software is associated with the SUNSET statistical software. However, such a use of SYLVIA requires a large number of runs and a significant statistical analysis what is not always compatible to the requirements of safety assessments in terms of deadlines. To overcome this difficulty, IRSN is currently developing expert systems based on SYLVIA result databases. This approach allows deriving the most likely diagnosis or prognosis in a very short time, but also deriving a more complex form of reasoning intertwining prognostic and diagnostic inferences. These expert systems are based on the Bayesian Belief Network methodology and consist in two steps: first, a large database obtained from SYLVIA runs allows the estimation of conditional probability tables; then, a message passing algorithm is used to dynamically exploit this database. The illustrating example is based on the study of the behavior of the final level of aerosol filtration in nuclear facilities, in a fire situation. The holding of the final level of filtration is conditioned by the thermal and mechanical stresses experienced by high-efficiency particulate air filters. The database of the expert system SEVEN is built from the results of ten million calculations performed with the SYLVIA software. This example illustrates how an expert system can be used as a decision support tool for fire safety analysis in the nuclear area. Expert systems represent a new generation of calculation tools in the field of probabilistic fire simulation and contribute to building the enhanced expertise of tomorrow.