In the process of innovative development of traditional manufacturing enterprises, the increasingly fierce competition among enterprises and the limitations of internal resources make it impossible for enterprises to adapt to the rapidly developing market demand. In view of this, this paper studies a new method for selecting innovation partners of traditional manufacturing enterprises from the perspective of resource complementarity. First, a partner selection index system is established from four dimensions: knowledge resources, technological resources, market resources, and environmental resources, and a method for determining the weights of evaluation indicators based on rough sets is constructed to fully reflect the true perception of the evaluation decision- making group and improve the rationality of determining indicator weights. Second, considering the fuzziness and uncertainty of evaluation information in the selection of candidate partners, an innovation partner selection model for traditional manufacturing enterprises integrating rough numbers and the technique for order preference by similarity to ideal solution (R-TOPSIS) is constructed. By incorporating the above-mentioned indicator weight information, the rough numbers of evaluation values and rough number ideal solutions of each candidate partner are calculated to achieve the optimal selection of partners. Finally, the selection of innovation partners of a traditional manufacturing enterprise is taken as a case study object to verify the feasibility and effectiveness of the proposed method.