Marine organisms exhibit great variation in reproductive modes, larval types, and other life-history traits that may have major evolutionary consequences. We measured local and regional patterns of genetic variation in corals along Australia's Great Barrier Reef to determine the relative contributions of sexual and asexual reproduction to recruitment and to infer levels of gene flow both locally (among adjacent sites, < 5 km apart) and regionally (among reefs separated by 500-1,200 km). We selected five common brooding species (Acropora cuneata, A. palifera, Pocillopora damicornis, Seriatopora hystrix, and Stylophora pistillata) and four broadcast spawners (Acropora hyacinthus, A. cytherea, A. millepora, and A. valida), which encompassed a wide range of larval types and potential dispersal capabilities. We found substantial genotypic diversity at local scales in six of the nine species (four brooders, two spawners). For these six, each local population displayed approximately the levels of multilocus genotypic diversity (Go) expected for outcrossed sexual reproduction (mean values of Go:Ge ranged from 0.85 to 1.02), although consistent single-locus heterozygous deficits indicate that inbreeding occurs at the scale of whole reefs. The remaining three species, the brooder S. hystrix and the spawners A. valida and A. millepora displayed significantly less multilocus genotypic diversity (Go) than was expected for outcrossed sexual reproduction (Ge) within each of several sites. Acropora valida and A. millepora showed evidence of extensive localized asexual replication: (1) a small number of multilocus (clonal) genotypes were numerically dominant within some sites (Go:Ge values were as low as 0.17 and 0.20): (2) single-locus genotype frequencies were characterized by both excesses and deficits of heterozygotes (cf. Hardy-Weinberg expectations), and (3) significant linkage disequilibria occurred. For the brooding S. hystrix Go:Ge values were also low within each of four sites (x = 0.48). However, this result most likely reflects the highly restricted dispersal of gametes or larvae, because levels of genetic variation among sites within reefs were extremely high (FSR = 0.28). For all species, we detected considerable genetic subdivision among sites within each reef (high FSR-values), and we infer that larval dispersal is surprisingly limited (i.e., Nem among sites ranging from 0.6 to 3.3 migrants per generation), even in species that have relatively long planktonic durations. Nevertheless, our estimates of allelic variation among reefs (FRT) also imply that for all four broadcast spawning species and three of the brooders, larval dispersal is sufficient to maintain moderate to high levels of gene flow along the entire Great Barrier Reef (i.e., Nem among reefs ranged from 5 to 31). In contrast, widespread populations of S. hystrix and S. pistilata (the two remaining brooders) are relatively weakly connected (Nem among reefs was 1.4 and 2.5, respectively). We conclude that most recruitment by corals is very local, particularly in brooders, but that enough propagules are widely dispersed to ensure that both broadcast spawning and brooding species form vast effectively panmictic populations on the Great Barrier Reef.