The neuro fuzzy system based on two independent structures is described, the first a neuro-observer system developed by use of dynamical neural networks, and the second as the control system based on fuzzy logic system. These structures are described by independent way and their properties are analyzed. Besides, the neuro-fuzzy system performance is proved by the application to the Bergman th blood Insulin-Glucose interaction model, the simulations show the neuro-fuzzy output as the insulin infusor output (insulin concentration), the glucose concentration estimated state is also described, as well as the inferential rules and the membership functions in the fuzzy. Copyright © 2005 IFAC.