In recent years, the notion of resilience has grown into an important concept for both scholars and practitioners working on disasters. This evolution reflects a growing interest from diverse disciplines in a holistic understanding of complex systems, including how societies interact with their environment. This new lens offers an opportunity to focus on communities’ ability to prepare for and adapt to the challenges posed by natural hazards, and the mechanism they have developed to cope and adapt to threats. This is important because repeated stresses and shocks still cause serious damages to communities across the world, despite efforts to better prepare for disasters. Scholars from a variety of disciplines have developed resilience frameworks both to guide macro-level policy decisions about where to invest in preparedness and to measure which systems perform best in limiting losses from disasters and ensuring rapid recovery. Yet there are competing conceptions of what resilience encompasses and how best to measure it. While there is a significant amount of scholarship produced on resilience, the lack of a shared understanding of its conceptual boundaries and means of measurement make it difficult to demonstrate the results or impact of resilience programs. If resilience is to emerge as a concept capable of aiding decision-makers in identifying socio-geographical areas of vulnerability and improving preparedness, then scholars and practitioners need to adopt a common lexicon on the different elements of the concept and harmonize understandings of the relationships amongst them and means of measuring them. This article reviews the origins and evolution of resilience as an interdisciplinary, conceptual umbrella term for efforts by different disciplines to tackle complex problems arising from more frequent natural disasters. It concludes that resilience is a useful concept for bridging different academic disciplines focused on this complex problem set, while acknowledging that specific measures of resilience will differ as different units and levels of analysis are employed to measure disparate research questions.