This paper proposes to use granular bodies of evidence to model uncertainty of our world. One of the advantages of granular information is that it incorporates fuzzy information and probabilities, which also can be represented with the help of fuzzy values. Moreover, these paradigms enable one to use natural language to describe the problem, which facilitates modelling of the decision to be made. However, generally fuzzy logic and probability theory is not used together to deal with uncertainty. We show some of the advantages of using two of these paradigms together and compare our results to the traditional approaches towards the modelling of uncertainty. Special attention is paid to the risk analysis, as it is closely related to that of uncertainty.