Analyses of risks related to ground and underground issues (surface collapses, subsidence, rockfalls, etc.) are generally undertaken in a strong context of uncertainty. However, tools that are available for geotechnical expert to carry out his study suffer today from not being able to really seize this context of uncertainty. This work takes firstly stock of the notion of uncertainty in risk analyses. It provides a definition and a typology of uncertainty that can be concretely used by the expert. For each of the defined classes, methods allowing an operational treatment of uncertainties are presented, either extracted from a literature survey or developed in the framework of the Thesis. The use of probabilities as an expertise-help tool is secondly examined. A discussion, which distinguishes between frequentist and epistemic interpretations of probabilities, is proposed to evaluate the benefits and implications that probabilities can have towards the practical elaboration of a risk analysis. This study is mainly dedicated to the field expert in charge of the analysis. Numerous concrete examples (analysis of surface stability above an underground coal mine, sinkhole development analysis over an underground gypsum mine, elaboration of a Mining Risk Prevention Plan) are therefore provided both to present the main results of the work and to illustrate the adaptability of the tools being introduced to the current methodologies of analysis. They also aim at highlighting the fact that the treatment of uncertainties and the use of probabilities in risk analyses can facilitate the process of expertise and allow a better communication between the various actors of risk management