Prior research into the link between new product development and market segmentation has focused on two main approaches: (1) design, segment, and do limited competitive evaluation; and (2) segment first, design second. This paper proposes a third approach, which is to simultaneously design, perform segmentation according to benefit and to evaluate against competitive designs. This research uses a benefit segmentation technique based on conjoint analysis (or other techniques that relate product attributes to consumer utility) in which the segments emerge simultaneously with the design based on certain design principles or "strategies." Herein a method is proposed to narrow down the many possible feasible designs (combinations of attributes) to a finite set and to examine the appeal of each design. Five distinct design strategies are proposed for modeling and studying competitive reaction. These include "traditional" ones such as differentiation and new ones whose fringe customers have high utility. The paper shows that these five strategies are adequate for modeling competitive reaction using simulation. Another contribution of the paper is the way competitive reaction is modeled. In generating and evaluating a design the desire herein is also to assess the defensibility of the design and include it in the evaluation criteria. These issues are addressed by decomposing the solution procedure into two phases. In the first phase, different optimal designs are created based on predefined product development strategies. In the second, these optimal designs are compared against one another with regard to market share and potential to secure market leadership. This work shows that the nature of competition as well as the variability of customer preferences are critical to how a strategy performs. This process uncovers a surprisingly robust design strategy-developing attributes such that a "lower fringe" is most satisfied-that even achieves market dominance under certain conditions. This methodology is also applied to partworth data on refrigerators, which provides a concrete example of the concepts and demonstrates results consistent with the propositions developed earlier in the paper.