A process plan describes the steps necessary to manufacture a product. Because creating a process plan is a time-comsuming activity, researchers have developed methods that reduce the time and effort required. The hybrid process planning approach builds on the strengths of the variant and generative process planning approaches while overcoming their individual drawbacks. This variant approach identifies useful process plans from a database that contains existing designs and their process plans. The retrieved plans are then merged and modified as necessary to generate a plan for the new design. To find these designs and process plans, one must define a design similarity measure that evaluates the similarity between an existing design and the new design. The hypothesis of the variant planning approach is that similar designs have similar process plans. However, existing design similarity measures are not plan-based. Consequently, the retrieved process plans may not be useful when building the new design︳ process plan. This thesis presents a systematic approach to formulate a precise and consistent plan-based design similarity measure that ensures that designs are classified as similar if and only if they have similar process plans. We do not assume a priori that similar designs have similar process plans. The proposed approach identifies functions that map design parameters to critical process planning attributes. We assume that there is a one-to-one correspondence between the designs and the corresponding process plans. Mapping functions identify those design parameters that must be similar for critical process plan attributes to be similar. Design similarity is thus a function of process plan similarity. This plan-based design similarity measure ensure that similar designs will have similar process plans. The generality of the proposed design similarity approach is validated by developing hierarchical and continuous design similarity measures using this approach.