We tested the hypothesis for several Caribbean reef fish species that there is no difference in nursery function among mangrove, seagrass and shallow reef habitat as measured by: (a) patterns of juvenile and adult density, (b) assemblage composition, and (c) relative predation rates. Results indicated that although some mangrove and seagrass sites showed characteristics of nursery habitats, this pattern was weak. While almost half of our mangrove and seagrass sites appeared to hold higher proportions of juvenile fish (all species pooled) than did reef sites, this pattern was significant in only two cases. In addition, only four of the six most abundant and commercially important species (Haemulon flavolineatum, Haemulon sciurus, Lutjanus apodus, Lutjanus mahogoni, Scarus iserti, and Sparisoma aurofrenatum) showed patterns of higher proportions of juvenile fish in mangrove and/or seagrass habitat(s) relative to coral reefs, and were limited to four of nine sites. Faunal similarity between reef and either mangrove or seagrass habitats was low, suggesting little, if any exchange between them. Finally, although relative risk of predation was lower in mangrove/seagrass than in reef habitats, variance in rates was substantial suggesting that not all mangrove/ seagrass habitats function equivalently. Specifically, relative risk varied between morning and afternoon, and between sites of similar habitat, yet varied little, in some cases, between habitats (mangrove/seagrass vs. coral reefs). Consequently, our results caution against generalizations that all mangrove and seagrass habitats have nursery function.