Obstructive sleep apnoea (OSA) is a common disorder characterized by repetitive closure of the upper airway during sleep and associated with significant adverse health effects including hypertension, heart disease and stroke. Current treatment with continuous positive airway pressure (CPAP) is highly effective but reduced compliance levels have resulted in suboptimal outcomes. Oral appliances such as mandibular advancement splints (MAS) are an alternative treatment and have potential advantages including greater patient compliance, comfort and portability. Although they have been shown to be successful across all categories of OSA severity, overall they are less effective than CPAP. A key limitation to its more widespread use has been the inability to predict which patients will be a treatment success. Prediction of treatment outcome would greatly enhance both MAS utilization and overall OSA management. However, little is known about the mechanisms of action of MAS therapy and a more detailed understanding is likely to improve patient selection and outcome. The aim of this thesis is to improve the prediction of treatment outcome through improved understanding of the mechanisms and site(s) of action of MAS therapy during sleep, through extrapolating this knowledge into daytime prediction tests and by developing prediction equations which can be tested prospectively. The work in this thesis presents novel ideas and findings. It is the first to examine and find that MAS therapy improves upper airway collapsibility during sleep. The site(s) of upper airway collapse was also examined and found to predict treatment outcome. Primary oropharyngeal collapse during sleep predicted treatment success and this was extrapolated into a simple daytime test hypothesized to reflect oropharyngeal function. These primary oropharyngeal collapsers were found to have characteristic awake flow-volume curves and this was then studied prospectively. Cephalometric X-rays and anthropomorphic measurements were also evaluated to formulate prediction equations for treatment outcome with MAS. These new findings together with their implications for clinical practice and future research are then summarized. It is concluded, however, that although many advancements have been made, the mechanisms of MAS action and prediction of treatment outcome remain incompletely understood reflecting the complex pathophysiology of the upper airway.