The task of design evaluation in conceptual design is important, since a poor selection of design concept can rarely be compensated at later design stages. At that stage, it is difficult to determine the 'best' design alternatives, because information collected is too subjective or too incomplete to make a judgment. In this paper, the imprecise preference structure of various alternatives is modelled with the fuzzy preference relation. Depending on the degree of completeness of design information, three preference modes based on the outranking approach are developed to discriminate the nondominance set from a set of alternatives for further development. The alternatives are removed from further consideration only if there is sufficient evidence to believe that they are outranked by other alternatives. A valve selection problem is used to illustrate the concept. The developed approach is more applicable in the imprecise and uncertain design environment.