Countries need to assess the expected risk due to natural hazards as a permanent activity in their financial plan; otherwise, they will experience a lack in the information required by the application of disaster risk reduction policies. In this article, a risk assessment methodology is proposed that uses, in one hand, empiric estimations of loss, based on information available in local disaster data bases, allowing to estimate losses due to small events; on the ot her hand, it uses probabilistic evaluations to estimate loss for greater or even catastrophic events for which information is not available due the lack of historical data. A “hybrid” loss exceedance curve, which represents the disaster risk in a proper and complete way, is thus determined. This curve merges two components: the corresponding to small and moderate losses, calculated by using an inductive and retrospective analysis, and the corresponding to extreme losses, calculated by using a deductive and prospective analysis. Applications of this new probabilistic risk assessment technique are given in this article for three countries.