Marine protected areas (MPAs) have proliferated in the past decades to protect biodiversity and sustain fisheries. However, most of the MPA networks have been designed without taking into account a critical factor: the larval dispersal patterns of populations within and outside the reserves. The scale and predictability of larval dispersal, however, remain unknown due to the difficulty of measuring dispersal when larvae are minute (~ cm) compared to the potential scale of dispersal (~ km). Nevertheless, genetic approaches can now be used to make estimates of larval dispersal. The following thesis describes self-recruitment and connectivity patterns of a coral reef fish species (Centropyge bicolor) in Kimbe Bay, Papua New Guinea. To do this, microsatellite markers were developed to evaluate fine-scale genetics and recruit assignment via genetic parentage analysis. In this method, offspring are assigned to potential parents, so that larval dispersal distances can then be inferred for each individual larvae. From a total of 255 adults and 426 juveniles collected only 2 parentoffspring pairs were assigned, representing less than 1% self-recruitment. Previous data from the same study system showed that both Chaetodon vagagundus and Amphiprion percula have consistent high self-recuitment rates (~ 60%), despite having contrasting life history traits. Since C. bicolor and C. vagabundus have similar characteristics (e.g. reproductive mode, pelagic larval duration), comparable results were expected. On the contrary, the results of this study showed that dispersal patterns cannot be generalized across species. Hence the importance of studying different species and seascapes to better understand the patterns of larval dispersal. This, in turn, will be essential to improve the design and implementation of MPAs as conservation and management tools.