The risk of severe and frequent injury associated with active pursuits in mountain areas is acknowledged. In France, the knowledge of such accident facts and causes seems perfectible. The first part of this presentation, rooted in the social construction of risks and based on 50 interviews, will consist of an analysis of this pessimistic statement: corporative, institutional, methodological and judicial stakes lead to a fragmented picture of accidentology and to the poor availability of data. The second part of the communication will investigate the perspectives offered by the study of not only accidents, but also near-miss sequences. This dedramatized way to improve knowledge of accident sequences consists in grasping the preventive value of problematic situations, which could have escalated to an accident but where worse outcomes have been avoided. Moreover, incidents, much more frequent than accidents, provide a greater pool of data to comprehend accidentology from. An international overview of existing accident and near-miss reporting systems will be presented. Although they provide interesting insights, most of them lack detailed information on the situations to enable thorough analysis of the data, and thus prevention improvement. We created a comprehensive accident and incident reporting system to allow in-depth, multi-factors and system-based analysis of reports. This applied research constitutes, in itself, a social innovation, since it currently resulted in the implementation of an online, open-access, participative platform enhancing the preventive use of gathered experience feedbacks. This process will be described in sociological terms, through such concepts as network dynamics, compromises, enrolment, and appropriation.