This paper proposes a framework based on the Dempster-Shafer Theory of Evidence (DSTE), Possibility Theory (PT) and Fuzzy Random Variables (FRVs) to represent and propagate the epistemic uncertainty in the parameters of the component degradation-failure models used to evaluate the performance of maintenance policies. An example of application is given with reference to a check valve of a turbo-pump lubricating system in a Nuclear Power Plant, which is affected by fatigue and overtakes condition-based maintenance interventions.