In product development, statistical experimentation techniques can help to improve product quality so as to meet or even exceed customers' requirements. To integrate proper statistical techniques effectively and sustainably into new product development and innovation processes, it is important to train product development engineers who are usually experts in their respective engineering domain. In this paper, a problem-based learning approach integrated within a Design for Six Sigma (DFSS) training program is described. It is illustrated that instead of relying on ad hoc implementation, statistical experimentation introduced through the DFSS methodology would enable product development engineers to appreciate its usefulness within a broader framework.