Abstract : The Business Intelligence (BI) is defined by the technologies developing and presenting the data in order to make decision. BI gives to the leaders a visibility on the performance of their company in order to improve their capacity in reacting more quickly than its competitors to new opportunities or to the risks of the market. The BI is based on a specific information system called Decisional Information System (DIS), in opposition to the transactional information systems. The DIS comprises several components which were summarized, formerly, in a data warehouse. A data warehouse is a collection of integrated and historized data that are used for strategic decision making by means of analytical techniques. The major part of the existing tools for the development of the data warehouses is focused on the structure of data storage. The interest is mainly carried to the definition of “star” models or “snow flake” models and the integration of data coming from heterogeneous sources, more than wondering about the requirements of the decision makers. Considering the requirements of the decision makers produces a DIS offering a better reliability in decision making. In addition, few requirement-driven approaches are proposed for DIS design. The principal goal is not to know “where” the data must be stored but “how” they should be structured and “why” they are necessary. The “Why” is, often, not known and even less not attached to the “what”. However, taking into account the “why” would make it possible to justify the choice of ‘right necessary' operational information. Consequently, the main research question of this thesis is “how to specify the requirements of the decision makers for a decisional information system which provides the most adequate information”. The main contribution of this thesis is a method of requirement engineering adapted to a decisional environment. This method comprises models of products and a process which guides the elicitation of the requirements and the design of the SID. This research contributed also to the definition of problems directly related to the needs of the industrial world and the research of the scientific community, its validation and its resolution by the proposal of a methodological approach and four meta-models, four models and two complementary ones. The suggested method is named CADWE (Computer Aided Data Warehouse Engineering). It exploits various sources to discover the requirements of a DIS: strategic goals strategic objectives of the organization and tactic objectives of the decision makers, ata of operational existing systems and a base of intentional and operational reusable components. From the whole of these resources, the process generates the multidimensional schema. The process of CADWE method has four steps which are (1) to identify the strategic goals, (2) to express the objectives, (3) to discover the informational requirements and (4) to define the multidimensional schema. This process reveals that various types of requirements, attached to a DIS, can be listed. The requirements of the DIS can be of four types: Strategic requirements related to the strategy of the organization, tactical requirements associated to a viewpoint, the operational requirements related to required information and the system requirements associated to the targeted system. These four types of requirements are indexed in two levels: The intentional level and the operational level. Lastly, the suggested method is illustrated by means of a real case study.