Type 2 diabetes is a complex chronic metabolic disorder characterized by hyperglycemia associated with a relative deficiency of insulin secretion and a reduced response of target tissues to insulin. Considerable efforts have been put into the development of models describing the glucose-insulin system. The best known is Bergman’s “minimal” model for glucose, which is estimating glucose concentrations using fixed insulin concentrations as input. However, due to the involved feedback mechanisms, simultaneous modeling of both entities would be advantageous. This is particularly relevant if the model is intended to be used as a predictive tool. The mechanism-based glucose-insulin model presented in this thesis is able to simultaneously describe glucose and insulin profiles following a wide variety of clinical provocation experiments, such as intravenous and oral glucose tolerance tests, clamp studies and sequential meal tests over 24 hours. It consists of sub-models for glucose, labeled glucose and insulin kinetics. It also incorporates control mechanisms for the regulation of glucose production, insulin secretion, and glucose uptake. Simultaneous analysis of all data by nonlinear mixed effect modeling was performed in NONMEM. Even if this model is a crude representation of a complex physiological system, its ability to represent the main processes of this system was established by identifying: 1) the difference in insulin secretion and insulin sensitivity between healthy volunteers and type 2 diabetics, 2) the action of incretin hormones after oral administration of glucose, 3) the circadian variation of insulin secretion and 4) the correct mechanism of action of a glucokinase activator, a new oral antidiabetic compound acting on both the pancreas and the liver. These promising results represent a proof of concept of a mechanistic drug-disease model that could play an important role in the clinical development of anti-diabetic drugs.