Increasingly, the success of management interventions aimed at biodiversity conservation are viewed as being dependent on the 'resilience' of the system. Although the term 'resilience' is increasingly used by policy makers and environmental managers, the concept of 'resilience' remains vague, varied and difficult to quantify. Here we clarify what this concept means from an ecological perspective, how it can be measured, and used to predict and manage for resilience. We argue that thresholds of disturbance are central to measuring resilience. Thresholds are important because they offer a means to quantify how much disturbance an ecosystem can absorb before switching to another state, and so indicate whether intervention might be necessary to promote the return of the predisturbance state. We distinguish between helpful and unhelpful resilience. Data to determine thresholds are not always available and so we consider the potential for indices of functional diversity to act as proxy measures of resilience. We also consider the contributions of connectivity and scale to resilience and how to incorporate these factors into management. We argue that linking thresholds to functional diversity indices may improve our ability to predict the resilience of ecosystems to future, potentially novel, disturbances according to their spatial and temporal scales of influence. Throughout, we provide guidance for the application of the resilience concept to ecosystem management. In doing so, we confirm its usefulness for improving biodiversity conservation in our rapidly changing world.