Abstract : The aim of the thesis is to evaluate the contribution of geodecisional approach for better analysing the problematic of natural disaster impacts on the French insurance market. An insurance company aims to get better insight on the natural disaster vulnerability. Two concerns have been identified: the first concern is “mutual” by reinforcing the efficiency of the public policy of natural disaster prevention, the second concern is “individual” by reinforcing the financial solidity of the company facing natural disaster event that could occur on their insurance portfolio. The offers in terms of tools and services to cover these requirements are limited for the French insurance market for two main reasons: first of all the French natural disaster indemnification system is based on a solidarity principle which has never encouraged the insurers to evaluate precisely the level of exposure of their clients to natural hazard, also, the second point is the heterogeneity of the spatial and non spatial data on the territory which has generated more complexity for the appropriation of the problematic by the French insurance companies. The thesis propose the “geodecisional approach” including a data modelling methodology (spatial multidimensional modelling) and technological components (for extraction, transformation, loading, restitution) in order to propose interface for exploring detailed and aggregated indicators combining spatial and non spatial analysis in a logical view according to the enterprise strategy. The idea is to adapt this approach to consolidate indicators for natural disaster vulnerability on the “mutual” and on the “individual” perspective. The different models are combining natural hazard maps like flood modellings, risk zoning maps in urbanism plans (“Plan de Prévention des Risques”) but also cadastral data and maps, geostatistical data about exposures and insurance data from an insurance company. The thesis will present the state of the art of the problematic and will propose a method for modelling and for implementing exploratory prototypes on case studies. The technology chosen to build the prototypes is Spatial OLAP developed at the Center of Reseach in Geomatics at Laval University in Quebec.