In the fuzzy decision table, the tolerance relation is built based on the ordered weighted averaging (OWA) operatorand, and OWA operator is a basic tool of information integration. Based on the OWA operator and its induced tolerance relation of cut sets, attribute reductions of the positive region, negative region and boundary region are discussed. Firstly, an attribute contribution degree distinguishing objects is defined by the weight of the OWA operator, to describe the reduction possibility of attributes. Consequently, a method of three-way attribute reduction based on the reduction possibility is proposed in the fuzzy decision table. The method reduces the search space and avoids reduction blindness. Finally, an example is analyzed to show the influencing causes of attribute reductions and the validity of the proposed method. The proposed measurement method of attribute contribution degree and the heuristic method of attribute reduction based on the three regions will reduce the searching space of attribute reduction, and thus they are benefit for applications of the fuzzy decision table.