Product design is an activity synthesized with art, technology and economics whose objective is to get a product satisfying the physical and psychological demands of human beings. However, the thought, behavior and language of human beings are full of fuzziness. The demands and degree of satisfaction that a person has with a product are usually represented with image words. Traditionally, the best design is subjectively selected from design alternatives based on designer's experience, delight, habit and inspiration when a lot of image information are faced. It is difficult to get a product satisfies, the customer’s demands with a decision model of that lacks of theoretical support. In this article, the design of a chair is taken as an example to construct a fuzzy decision model for a system with multi-level factors. Computerized questionnaires are used to get user’s psychological responses to each part of the chair, which are then converted, into the membership functions. The weighted mean and weighted generalized mean methods for fuzzy operations are used to quantify the fuzzy images. After comparing the membership grades of the design alternatives, the optimal design can be selected by the designer. With this decision model, not only can the error of a design decision be reduced, but the efficiency of the decision-making can also be improved. In addition, this decision model can further be used to design other more complicated products.