In this paper, we present an approach to query modeling that leverages the temporal distribution of documents in an initially retrieved set of documents.