Effective agitation management through sedative delivery is a fundamental activity in the intensive care unit (ICU), and is the basis for providing comfort and relief to the critically ill. The primary goal of sedation management in the ICU is to manage patient agitation, the sometimes violent activity that creates a risk to both patients and clinical staff. Recent statistics show that 71% of critically ill patients experience agitation for 58% of their stay in ICU, so the problem is significant in size and patients affected. In clinical practice, a lack of understanding of the underlying dynamics, combined with subjective agitation assessment tools, makes effective and consistent agitation management difficult. Quantitative models of agitation-sedation dynamics are developed to enhance understanding of the underlying system, and to create a platform enabling the development of sedative control protocols for improved agitation management. Compartmental pharmacokinetic models are employed to model the transport and elimination of drugs, and response surface modelling is used to represent the pharmacodynamic effect of the drugs on patient agitation. These physiologically-based models incorporate observations of critically ill patients, and are evaluated against recorded infusion data. The models are then used to evaluate different methods of sedation and patient agitation management. Constant infusion protocols, derivative-focused agitation feedback control protocols, and non-linear agitation feedback protocols are specifically investigated. The results identify physiological effect saturation as a severe limitation on the capacity of sedative control protocols to reduce agitation. In spite of this restriction, significant improvements over current clinical practice are achieved through the use of very simple control protocols. These improvements represent significant benefits for patients and medical staff alike. However, the fundamental outcome of this thesis is that improved agitation management is not determined so much by the class of drug or drug dose, as by the specific drug administration strategy employed.