In high-value-added machining of complicated shapes, machining know-how is indispensable regarding workpiece clamping and fixing, selection of cutting tools and decision of cutting conditions. However, the process planning still depends on skillful experts to achieve complex-part machining. Therefore, in order to realize high efficient machining, it is necessary to formulate the implicit machining know-how owned by skillful experts and to obtain these tacit knowledges. This study focuses on rough machining of complicated shapes as the first step and proposes a method to logically decide workpiece shapes during a machining operation by using topology optimization.