Population connectivity, the extent to which geographically separated subpopulations exchange individuals and are demographically linked, is important to the scientific management of marine living resources. In theory, the design of a marine protected area, for example, depends on an explicit understanding of how dispersal of planktonic larvae affects metapopulation structure and dynamics (Botsford et al. 2001). In practice, for most marine metazoans with planktonic larvae, the mean and variance of the distances that larvae disperse are unobservable quantities, owing to the small sizes of larvae and the very large volumes through which they are distributed. Simulation of dispersal kernels with models that incorporate oceanography and limited aspects of larval biology and behaviour, coupled with field studies of larval distribution, abundance, and settlement, have provided the best available approaches to date for understanding connectivity of marine populations (Cowen et al. 2006). On the other hand, marine population connectivity has often been judged by spatial variation in the frequencies of alleles and genotypes, although the inherent limitations of this indirect approach to measuring larval dispersal have often been overlooked (Hedgecock et al. 2007). More recently, researchers have turned to genetic methods and highly polymorphic markers that can provide direct evidence of population connectivity in the form of parentage or relatedness of recruits (e.g. Jones et al. 2005). In this issue, Christie et al. (2010) provide a particularly elegant example, in which both indirect and novel direct genetic methods are used to determine the major ecological processes shaping dispersal patterns of larval bicolour damselfish Stegastes partitus, a common and widespread reef fish species in the Caribbean Basin (Fig. 1).