During the last years, Binary Decision Diagrams (BDDs) have attracted the attention of researchers for reliability assessment of complex and large scale systems because they reduce significantly the time and resource demand of reliability computations. In this paper, we present an original extension of BDDs for reliability assessment of systems in the presence of epistemic uncertainty. The proposed method computes the imprecise system reliability from the imprecise reliabilities of the components presented as intervals. This method computes the imprecise system reliability, much faster than the classical inclusion-exclusion method. The efficiency of our proposed method will be demonstrated using numerical examples of imprecise reliability computation of several system configurations in term of time consumed for system reliability computation