The principles of nonlinear dynamics have found applications in areas as diverse as meteorology, ecology, economics, and business management. In medicine, insights have been gained into the nature of physiological and pathologic processes, the organization of healthcare systems and in the design of diagnostic and therapeutic tools. This article attempts to further extend these principles to improve the everyday care of people with diabetes, particularly those with type I disease, aiming at tight glycemic control. A model is proposed that recognizes the core determinants of blood glucose as mutually dependent variables governed by nonlinear interactions, specifying a dynamic "phase space" through which the patient navigates. Although these determinants are greater in number than is often recognized, they are few enough that the distinction between random variation and low-dimensional chaos, which is central to nonlinear dynamics theory, becomes useful.