A decision method based on rough sets is proposed for multi-attribute decision making, in which the attribute values are in the form of hesitant fuzzy elements. Firstly, the discernibility matrix is acquired according to the given threshold and the similarity degree between attribute values and ideal points. Then, the attribute reduction is operated and the weight of core attribute is determined by the discernibility matrix. Finally, the weighted similarity degree of every alternative with ideal
points is displayed to rank all the alternatives. An example is given to illustrate the effectiveness of the proposed method.