It is an important subject how deal with the symptom's data, input data, to improve the accuracy and efficiency of the diagnostic algorithm in the medical decision support systems. In this paper, we described a method for the numerical value of symptoms to develop the medical decision support system that diagnose a large number of diseases, not one or two diseases. The numerical value of symptoms is realized by dividing the symptom into several parts in consideration of the following contents and setting up the specific value in each part. That is, the symptom is appeared in where, when the symptom is appeared, what about the symptom, How long the symptom is, etc. Then we decided the distance(similarity) between symptoms on the basis of the numerical value of symptoms. Determination of the distance between symptoms by numerical value enables estimating the symptoms of which disease is nearer from the symptoms of patient and obtaining the suitable diagnostic result. So, we established useful method of high accuracy to diagnose various diseases revealed by different symptoms.