6σ has proved numerous times that it can help any business to deliver products to meet or even exceed customer expectations. DMAIC (Define – Measure – Analyze – Improve - Control) is one of the Six Sigma methodologies used for a 6σ project that aims to improve and control quality of exiting products. Aiming at the drawbacks existed in the problem selection in definition stage, considering the fuzziness and uncertainty of problem analyzing, a method integrating fuzzy QFD (Quality Function Deployment), frequency-cost FMEA (Failure Mode Effect Analysis) and gray relation analysis is presented, in which the customer requirements, occurring frequencies and the relative costs of the problems are taken into account comprehensively. Thus the key quality characteristics or process features to be improved are determined, which can provide decision support for key problem selection. An example is given to illustrate the effectiveness of the proposed method.