The main purpose of this paper is to answer this question: How to explain FMEA difficulties and how to improve FMEA methodology in order to be able to design a better risk management strategy in NPD? First, this research will discuss the FMEA procedure, its history, its main concepts and its current weakness. Then, we propose to highlight that the limits of FMEA procedure can be linked to the limits of a problem-solving paradigm. We reveal that current vision of engineering design, and particularly FMEA procedure, is based on problem solving perspective, which is restricted by the “bounded-rationality” model. From the concept of “expandable rationality” we propose to revisit the theoretical framework of the FMEA to explain why FMEA cannot be reduced to problem solving. Finally, we argue that FMEA procedure is a full design activity and we propose to extend the initial FMEA methodology, by using CK design theory. The research was carried out in STMicroelectronics manufacturing department, more specifically in the Engineering Competences Center located in Crolles (France).