Modularity is the key to improve the cost-variety trade-off in product family development. This paper presents a methodology for identifying the constituent modules of product family. The methodology includes the following principles: (1) identify and isolate the individualized components into one module so that the product’s differentiation point is postponed; (2) identify and isolate the components with high possibility of being changed in the future into one module in order to improve the stability and commonality in the product family; (3) improve the functional independency of the modules so as to support the module configuration process and the functional extension of the product family; and (4) improve the structural independency of the modules so as to achieve higher efficiency in the module manufacturing process. These four principles are incorporated into a mathematical model and therefore the module identification problem is translated into a multi-objective combinatorial optimization problem. The problem is solved using genetic algorithm (GA). A case study is carried out on a gear reducer. Sensitivity analyses show that relative weights of each principle and different initial numbers of modules result in different module schemes.