A challengeable environment of semiconductor industry can be characterized by uncertain demands, the long manufacturing lead-time, the expensive capacity cost and relatively short life cycle, all of which means significant capital and big risk. It is critical to design a simple and useful approach for yield planning. Since system dynamics (SD) is an approach to understanding the behaviors of complex systems, this paper presents a system dynamic model to determine the optimal yield of the semiconductor manufacturing system. In the paper, we introduce customer lifetime value (CLV) as a measure of revenues to achieve the long-term profits, rather than decisions on maximum short-term profits. The results show that SD is an effective simulation approach, and several interesting and useful insights are found.