Correct identification of closely related species is important for reliable measures of gene flow. Incorrectly lumping individuals of different species together has been shown to over- or underestimate population differentiation, but examples highlighting when these different results are observed in empirical datasets are rare. Using 199 single nucleotide polymorphisms, we assigned 768 individuals in the Acropora hyacinthus and A. cytherea morphospecies complexes to each of eight previously identified cryptic genetic species and measured intraspecific genetic differentiation across three geographic scales (within reefs, among reefs within an archipelago, and among Pacific archipelagos). We then compared these calculations to estimated genetic differentiation at each scale with all cryptic genetic species mixed as if we could not tell them apart. At the reef scale, correct genetic species identification yielded lower FST estimates and fewer significant comparisons than when species were mixed, raising estimates of short-scale gene flow. In contrast, correct genetic species identification at large spatial scales yielded higher FST measurements than mixed-species comparisons, lowering estimates of long-term gene flow among archipelagos. A meta-analysis of published population genetic studies in corals found similar results: FST estimates at small spatial scales were lower and significance was found less often in studies that controlled for cryptic species. Our results and these prior datasets controlling for cryptic species suggest that genetic differentiation among local reefs may be lower than what has generally been reported in the literature. Not properly controlling for cryptic species structure can bias population genetic analyses in different directions across spatial scales, and this has important implications for conservation strategies that rely on these estimates.