Recovery of the Caribbean long-spined sea urchin Diadema antillarum from the pathogen-induced mass mortality event of 1983 has been slow and variable. Multiple studies indicate that post-mortality population densities are greatest in sheltered lagoon and back reef habitats, and that recovery is limited on forereefs where pre-mortality densities were also high (Miller R. J. et al., 2003;Van der Hal, 2005;Debrot and Nagelkerken, 2006;Steiner and Williams, 2006;Vermeij et al., 2010). We carried out detailed surveys of a recovering population of D. antillarum in Curacao to document an ongoing influence of habitat type on population density, and to explore what ecosystem characteristics might drive observed recovery patterns. We quantified D. antillarum population density, larval settlement rates, herbivorous fish biomass, algal turfs, macroalgae, and predatory fish biomass in sites with and without structural complexity on reefs in sheltered lagoon and backreef locations, and on the forereef, using an orthogonal survey design. D. antillarum densities were greatest in sheltered lagoon and backreef locations, with high structural complexity. However, densities were significantly lower than those reported prior to the mass mortality event. Larval settlement rates were greater in sheltered lagoon and backreef locations but did not fully account for observed patterns in urchin density. Sheltered lagoon and backreef habitats had more turf algae and fewer herbivorous fish than forereef habitats. Predator abundance was generally low and did not differ significantly among habitat types. In light of our results, we hypothesize that patterns of D. antillarum recovery in Curacao are consistent with dynamics expected in the presence of density-dependent habitat selection. More algal resources and fewer herbivorous fish provide a mechanism for increased fitness and habitat suitability in sheltered, lagoon, and backreef habitats. Forereefs, by contrast, offer reduced habitat suitability and, we hypothesize, will only be re-colonized once the carrying capacity of more favorable habitats is exceeded. We propose a number of testable predictions to explore this theory further, and advocate for including sheltered lagoon and backreef habitats in D. antillarum population surveys to monitor recovery, as well as protecting populations found in these habitats where necessary. We predict that when current Rogers and Lorenzen Density Dependent Habitat Selection in Diadema antillarum populations return to pre-mortality densities, range expansion may allow for a return of D. antillarum to Caribbean forereefs.