We investigated the effects of mangrove cutting on fish assemblages in Twin Cays, Belize, in two habitat types. We conducted visual censuses at two sites in adjoining undisturbed/disturbed (30%–70% of shoreline fringe removed) sub-tidal fringing Rhizophora mangle Linnaeus, 1753. Observers recorded significantly more species and individuals in undisturbed sites, especially among smaller, schooling species (e.g., atherinids, clupeids), where densities were up to 200 times greater in undisturbed habitat. Multivariate analyses showed distinct species assemblages between habitats at both sites. In addition, extensive trapping with wire minnow traps within the intertidal zone in both undisturbed and disturbed fringing and transition (landward) mangrove forests was conducted. Catch rates were low: 638 individuals from 24 species over 563 trap-nights. Trap data, however, indicated that mangrove disturbance had minimal effect on species composition in either forest type (fringe/transition). Different results from the two methods (and habitat types) may be explained by two factors: (1) a larger and more detectable species pool in the subtidal habitat, with visual "access" to all species, and (2) the selective nature of trapping. Our data indicate that even partial clearing of shoreline and more landward mangroves can have a significant impact on local fish assemblages.