The information sets are recasted as information granules by using the fuzzy equivalence relations. The proposed information granules can be visualized as the entropies corresponding to the information source values, drawn together by their similarity. The information granules are then further used to articulate rough information sets (RIS). The RIS are particularly useful in the approximation of decision concepts in terms of entropies (information). The usefulness of the rough information sets is demonstrated through a case-study.