Supplier selection has emerged as an irreplaceable element in the supply chain management. Traditional multi-criteria decision-making view to the supplier selection is confronted with some drawbacks such as the aggregation of different criteria. In this paper, fuzzy preference relation is used to express the relative preference of pairwise suppliers, which are based on the linguistic values given by domain experts. Ordered weighted averaging operator is applied to develop two types of coefficients, preference coefficient and aversion coefficient, respectively, to express the preference degree and aversion degree for each supplier. Finally, the best supplier is determined according to the measure of preference coefficient. The effectiveness of the proposed method has been demonstrated by an illustrative example.