In this paper, among multi-criteria decision making (MCDM) procedures, the ELECTRE(Elimination and (Et) Choice Translating Reality)-III is revised to solve the supplier selection problem where both a criteria evaluation matrix and a decision matrix are linguistically evaluated by a group of decision makers. Final weights are obtained through subjective weights from criteria matrix and weights of the entropy from decision matrix. In addition, the fuzzy theory for dealing with ambiguity of linguistic variables is incorporated to develop the new procedure of the ELECTRE-III. The proposed procedure is tested through the numerical example which was used in the previous research with the fuzzy TOPSIS (technique for order preference by similarity to ideal solution). The results of the proposed procedure are compared to validate the outranking orders of suppliers. On the other hand, sensitivity analyses for the five cases with different subjective weights are carried to investigate the changes in the outranking orders of suppliers. In conclusion, the fuzzy ELECTRE-III procedure with entropy weights considering subjective judgement seems to be useful in the supplier selection problem.