The primary building block of coral reef ecosystems, whether an isolated mid-ocean atoll or a complex system of thousands of reefs, are the individual coral colonies that combine to form the reef structures which host coral communities. The state of a coral reef community (i.e. percentage coral coverage, dominant growth forms and size classes), or any other ecosystem, at any given time is the result of interactions between a range of disturbances and routine ecological processes that operate across a continuum of spatial and temporal scales (Gunderson et al 2002). The biological (living coral colonies) and structural (dead coral framework and rubble) legacies left behind in coral communities after a disturbance influence both the vulnerability of those communities to further disturbance and their ability to maintain current community structure: ecological resilience (Nystrom and Folke 2001). While some disturbances, particularly damage from tropical cyclone waves, can affect large areas with a single event, damage is invariably patchy in distribution because vulnerability varies at many scales, from individual coral colonies upwards. Thus, disturbance regimes need to be understood at regional scales (100s of km) to capture their full extent, and at very local scales (10s of m) to assess their potential impacts. Doing so has proved difficult because direct observations of damage are typically only possible at very local scales, at which high variability obscures regional patterns (Schneider 2001). A few studies have attempted to characterise the dynamics of large, infrequent disturbances across broad regions – for example, periodic forest fires and volcanic eruptions – but only for land-based ecosystems (Turner and Dale 1998). This paper describes the first such characterisation for a large marine ecosystem, which examines tropical cyclone disturbance across the Great Barrier Reef (GBR) over 35 years (1969-2003).