To analyse impacts and costs of wind storms on infrastructures, 24 major storms from the 1990-2010 period that occurred in the canton of Vaud are selected. Fine resolution loss data from these events is provided by a local insurance industry. This PhD work is subdivided into several sections: the spatial distribution of the loss data from these 24 storms is first studied using Geographic Information Systems (GIS). A map of extreme wind speeds at a 50 meter resolution over Switzerland is then created using both Generalized Additive Models (GAMs) and GIS tools; this wind map subsequently serves as input to apply storm loss models to the canton of Vaud. Regional Climate Models (RCMs) are finally used to reproduce these 24 storms, and furthermore to assess the potential climate change effects on future wind speeds in Switzerland and in Vaud. This thesis highlights the challenge of accurately estimating wind speeds over complex terrain and to precisely predict wind storm losses. However, simple assumptions and calculations indicate that economic losses related to wind storms are likely to drastically increase as a result of demographic growth and enhanced future wind speeds.