Disease-processes experienced by critically ill patients commonly cause unique pharmacokinetic (PK) changes, that may result in drug concentrations that are either subtherapeutic or toxic when dosing is based on PK data obtained from non-critically ill patients. Ensuring therapeutic dosing of beta-lactam antibiotics in critically ill patient is of increasing concern, given the difficult-to-predict PK in this population, frequent use of this antibiotic class in treating severe infections and the importance of optimal antibiotic therapy for reducing mortality rates. Therapeutic drug monitoring (TDM), a strategy frequently used in dosing drugs with complex PK, has demonstrated potential clinical advantages for improving dosing of beta-lactams in the critically ill. However, the practice of beta-lactam TDM and dose optimisation is still limited by a number of factors. Large-scale data on the adequacy of current dosing regimens and optimal PK/pharmacodynamic (PK/PD) targets associated with maximal clinical outcome is still limited. In particular, currently available data on betalactam TDM have not adequately quantified the potentially profound impact of hypoalbuminaemia, common among the critically ill, on the PK of the pharmacologically active unbound concentration of beta-lactams. This Thesis aims to describe the application of a beta-lactam TDM-guided dosing program in critically ill patients, with the utilization of directly-measured unbound drug concentration data. Specifically, this work will first characterise the impact of altered protein binding in critically ill patients on the accuracy of commonly used methods for calculation of unbound beta-lactam concentrations (from measured total concentrations). Secondly, this work will describe the achievement of PK/PD targets in relation to directly-measured unbound betalactam concentrations in critically ill patients receiving empiric dosed and TDM-guided betalactam antibiotic treatments. Finally, the association of beta-lactam concentrations with clinical outcome in critically ill patients with blood stream infections will also be characterised. This Thesis comprises of seven chapters. Chapter one introduces relevant PK/PD and betalactam dosing concepts for this Thesis. Chapter two includes a published review article that systemically analyses the current literature at the time on the risk of inadequate drug exposure in critically ill patients and V Publications included in this Thesis Wong G, Sime FB, Lipman J, Roberts JA. How do we use therapeutic drug monitoring to improve outcomes from severe infections in critically ill patients? BMC Infectious Diseases.