Violence against women is pandemic; globally 30% of women have experienced intimate partner violence. A review of literature indicated that even though intimate partner violence is a major public health issue, the health sector has been slow to respond, firstly by not identifying and responding to it in women's health consultations, and secondly by poor data collection. Emergency departments see a minority but important sector of women exposed to intimate partner violence: women most heavily abused by a partner. This thesis starts from a position, based on previous research in the field, that often when women attend an emergency department after an assault by their partner, their experience of intimate partner violence is 'missed'. Framed by the sociology of diagnosis, the focus of this thesis is the (mis)dassification of intimate partner violence in hospital-based emergency department heath systems. This thesis was positioned ontologically and epistemologically through a synthesis of critical realism and complexity theory and employed a mixed-method research design involving interviews with women victim/survivors of intimate partner violence, health practitioners, and clinical coders; a review of emergency department attendance records; and a survey of administrative health data. The claim made in this thesis is that, for hospital-based emergency department health systems, the best classification of intimate partner violence, in the form of physical assault, is 'assault by partner'. This claim is based on research findings presented in this thesis which indicated 'missed' intimate partner violence was a result of misclassification of 'intimate partner violence' into classifications that did not mobilize classification or intervention other than routine medical care. 'Assault by partner' was identified as best classification because there was no need for additional distinctions to be made: most patients in this study had experienced severe violence and suffered medium and high levels of injury. Furthermore, 'assault by partner' was proposed as best classification because of conceptual uncertainty and difficulty for health practitioners and clinical coders to classify 'domestic violence'. This thesis makes original contributions in the fields of sociology of violence against women, sociology of diagnosis, and health policy. From the research findings, recommendations for policy and practice to improve hospital-based emergency department system responses to intimate partner violence have been made.