Several models have been developed for decades. These models have an inherent drawback: for each blood glucose value, a different insulin infusion rate is needed to maintain constant blood glucose level. The objective is to review the elementary modelling to eliminate this contradiction with real life. A new model is derived for type-I diabetics. It has realistic asymptotic properties as there is one single constant insulin infusion rate, known as the basal rate, independent on the glycemia, and which ensures the equilibrium of any value of the glycemia in fasting periods as it is in real life. A standard identification algorithm is used to get the patients’ individual characteristics. Standard clinical data (CGM, injection, CHO uptake) were analyzed for a typical type-I diabetic outpatient. A long term fit over two days with constant parameters is realized. Moreover, the new model provides the computation of the tools for the functional insulin therapy like basal infusion rate, the insulin sensitivity factor (ISF) or insulin to carbo ration (I:C). This is a major outcome as it can be used for the education of type-I diabetics.