In medicine, diagnostic reasoning refers to the approaches used by physicians with the aim of achieving a medical diagnosis concerning a given patient. This paper presents a new approach of medical decision support systems. The proposed approach is based on the use of possibility theory as a global framework, including knowledge representation (as a possibilistic pair of measures: Necessity, Possibility); and, building a possibilistic medical knowledge base (to be exploited in order to make a diagnostic decision (classification of new medical cases). The efficiency validation of the proposed approach is conducted using an Endoscopic Knowledge and Case Base systems. Obtained results confirm that the proposed approach constitutes an efficient tool in terms of medical knowledge representation and possibilistic diagnostic reasoning.