Although the establishment of reserves is central to marine conservation, previous criteria for the selection of appropriate areas has often been based on historical, aesthetic or logistic factors, resulting in a network of marine reserves that may not effectively meet conservation objectives. Since the dispersal of larvae plays an integral role in determining whether reserves can sustain themselves, whether they can exchange larvae with other protected sites, or whether they can supplement surrounding exploited areas, effective reserve design requires consideration of the processes that may affect larval dispersal. Plankton tows were conducted monthly for 12 mo in a semi-enclosed marine reserve with long water retention time, in bays with low water retention, and along open coastline, to establish whether larval retention plays an important role in limiting dispersal and creating discrete communities. Significant spatial and temporal differences in larval assemblages were found, with the semi-enclosed reserve, bay, and coast areas showing consistent differences throughout the year. An ANOVA carried out on 11 species identified as differentiating between these areas supported the hypothesis that limited larval exchange was occurring between reserve and non-reserve areas, despite potentially large dispersal distances. Designating bays as reserve areas may provide a means of protecting various species with both long and short dispersal distances, but this must be balanced against the reduced likelihood of closed populations to recover from local catastrophes and the possibility of limited genetic differentiation and inbreeding.