Dealing with uncertainty introduces an increased level of complexity to reliability analysis problems. The uncertainties associated to reliability studies usually arise from the difficulty to account for incomplete or imprecise reliability data and complex failure dependencies. This paper introduces the Transferable Belief Model (TBM) to the reliability analysis so that epistemic uncertainties can be taken into account as well as aleatory uncertainties. Two approaches are used to represent failure dependencies of components: an implicit and an explicit approach. The TBM model is then compared to an intervalprobability model by highlighting the different characteristics of the results obtained