Three-way decision is a philosophy of thinking in three. It provides a simple and efficient solution for addressing complicated problems. In this paper, hesitant fuzzy sets and three-way decision methods are investigated by virtue of soft set theory. Value spaces and value cosets of hesitant fuzzy sets are defined. The notions of canonical soft sets, unit interval parameterized soft sets and derived hesitant fuzzy sets are presented to solve the mutual representation problems regarding hesitant fuzzy sets and soft sets. In addition, a generalized rough model based on hesitant fuzzy sets is established with the help of soft rough sets. Based on the given pre-decision set, soft upper approximations are calculated to determine the evaluation function. A novel hesitant fuzzy three-way decision method based on soft rough set is proposed. Finally, two numerical examples with related comparative analysis are given to verify the effectiveness and cogency of our method.