The importance of seagrass beds and mangroves as juvenile habitats for coral reef fishes is still debatable. We hypothesised that the configuration of these habitats in the marine landscape in relation to the coral reef would influence accessibility from the reef (both for fishes that use these habitats for shelter or feeding, and their predators). In combination with differences in habitat complexity (related to habitat type) this could influence the utilisation of these habitats by juvenile reef fishes, and connectivity with the coral reef. Using underwater visual census, we studied the use by juvenile reef fishes of various seagrass and mangrove habitats situated differently in relation to the coral reef, on the Caribbean island of Aruba. Additionally, we studied fish assemblages on various coral reefs differing in proximity to seagrass Thalassia testudinum and mangrove Rhizophora mangle habitats at the scale of the whole island. In the seagrass and mangrove habitats, habitat configuration in relation to the reef was related to the composition of the fish assemblage, species richness, fish density and fish size. This was consistent with our expectations related to accessibility of these habitats from the reef, and habitat complexity. Most fish species that use seagrasses and mangroves as juvenile habitats were absent from or showed reduced densities on coral reefs located far (> 9 km) from seagrass-mangrove habitats. At a smaller spatial scale, this effect was less clear. For some species, this lack of an effect at smaller spatial scales may be explained by local recruitment to the reef, whereas for other species (i.e. Haemulon sciurus, Lutjanus apodus, L. mahogoni and Scarus guacamaia) migration along the coast offers a more likely explanation. We suggest that the value of seagrass and mangrove habitats as a juvenile habitat should not be generalised a priori, since habitat configuration may interact with the degree of connectivity between seagrasses, mangroves and coral reefs.