Builders have always been concerned about the durability of construction materials and building components. Nowadays, powerful information technology tools offer an opportunity to approach the complexity of aging phenomena in an innovative way. This thesis presents a method and a simulation model designed to predict the aging of multi-layer building components. Integrated into a software tool, they take into account the nature and intensity of the external agents acting on a component. The tool creates possibilities for a variety of applications : it can guide users towards the most appropriate and durable construction systems, it can be helpful in decision making for building maintenance and it can provide service-life information for life-cycle analysis (LCA). The work is based on research undertaken between 1990 and 1997 by the Expert Service from the Construction and Conservation Laboratory (LCC) at the Ecole Polytechnique Fédérale de Lausanne (EPFL) in collaboration with the ARIA-laboratory of the Ecole Nationale Supérieure d'Architecture de Lyon (UMR CNRS n°694 MAP, Hervé Lequay) for the development of the Assistance to Maintenance of Buildings tool (AMB). The goal of the present work was to continue and extend the previous investigations. In a first step, the tool has been analysed and recreated within the Delphi development environment. New parameter recorder modules and graphical analysing modules have been added for better simulation monitoring and visualisation of chain reactions due to degradation processes. The model has also been adapted to allow stochastic simulation with the Monte-Carlo Method. This part of the research permits bringing out the structure of the simulation method's various aspects and concepts and working out its detailed description. In the report, global aspects of the prediction method are presented first. Basic concepts and principles, such as stress actions, material performance and element functions, are described. In a second part, the design of the computer model is presented. The aim is to show how the general concepts of the method can find an application. Various stress agents and specific material performances that participate in the aging process and lead to a decrease in performance are highlighted. Aging is simulated by the interaction between the materials and agents over time. Definitions of the failure criteria for a given building component are made by identification of its functions and assigning performance limits values. Two chapters are dedicated to the detailed description of the different stress agents and performances used in the model. These are followed by a selection of case studies that shows the working of the simulation and illustrates the various types of results that can be obtained. The conclusions reiterate that the prediction of construction material aging is a wide and complex subject. To be able to make not only qualitative but also quantitative durability predictions, further research must be undertaken in different domains. This is necessary in order to collect existing knowledge and to obtain missing information. As the presented method and model can easily integrate new data as it becomes available, they could play the role of a development platform for future investigations. This would allow a progressive increase in the reliability of the predictions.