Larval dispersal is a critical yet enigmatic process in the persistence and productivity of marine metapopulations. Empirical data on larval dispersal remain scarce, hindering the use of spatial management tools in efforts to sustain ocean biodiversity and fisheries. Here we document dispersal among subpopulations of clownfish (Amphiprion percula) and butterflyfish (Chaetodon vagabundus) from eight sites across a large seascape (10,000 km 2 ) in Papua New Guinea across 2 years. Dispersal of clownfish was consistent between years, with mean observed dispersal distances of 15 km and 10 km in 2009 and 2011, respectively. A Laplacian statistical distribution (the dispersal kernel) predicted a mean dispersal distance of 13-19 km, with 90% of settlement occurring within 31-43 km. Mean dispersal distances were considerably greater (43-64 km) for butterflyfish, with kernels declining only gradually from spawning locations. We demonstrate that dispersal can be measured on spatial scales sufficient to inform the design of and test the performance of marine reserve networks. R obust descriptions of larval dispersal are fundamental to studies of fish population dynamics 1,2 , fisheries management 3,4 and the design of reserve networks tasked with conserving ocean biodiversity 5,6 . Yet descriptions of larval dispersal patterns in ocean environments remain scarce. The combination of a pelagic larval phase that may last several days to many months and an ocean environment characterized by energetic diffusive and advective flows may allow passive larvae to disperse hundreds to thousands of kilometres from natal locations 7,8 . It has proved difficult, however, to verify directly how far fish larvae travel, because it is almost impossible to follow them as they disperse rapidly from spawning sites and are subject to high rates of natural mortality throughout the larval phase 9 . Our inability to describe the spatial extent of larval dispersal is problematic because our understanding of metapopulation dynamics relies on largely untested models that quantify where larvae arriving at a subpopulation originate from and where larvae spawned at each subpopulation eventually settle [10][11][12] . Moreover, to be of practical use, these data must be assembled on large enough scales for evaluating and optimizing spatial management strategies for fisheries or conservation 1,13 .Patches of reef habitat are frequently isolated from each other by deeper water that forms a barrier to adult movement, and so larval dispersal is likely to be a critical process in the persistence of many reef fish populations over demographic and evolutionary timescales 10,14 . Effective management of coral-reef seascapes is therefore particularly reliant on spatial tools to achieve conservation objectives. Although reef fish larvae clearly have the potential for long-distance movements 14 , there is increasing evidence that dispersal may be more limited than previously assumed 15,16 . The most compelling evidence of larvae returning to natal or nearby reefs has...