Coral reef ecosystems are based on coral-zooxanthellae symbiosis. During the initiation of symbiosis, majority of corals acquire their own zooxanthellae (specifically from the dinoflagellate genus Symbiodinium) from surrounding environments. The mechanisms underlying the initial establishment of symbiosis have attracted much interest, and numerous field and laboratory experiments have been conducted to elucidate this establishment. However, it is still unclear whether the host corals selectively or randomly acquire their symbionts from surrounding environments. To address this issue, we initially compared genetic compositions of Symbiodinium within naturally settled about 2-week-old Acropora coral juveniles (recruits) and those in the adjacent seawater as the potential symbiont source. We then performed infection tests using several types of Symbiodinium culture strains and apo-symbiotic (does not have Symbiodinium cells yet) Acropora coral larvae. Our field observations indicated apparent preference toward specific Symbiodinium genotypes (A1 and D1-4) within the recruits, despite a rich abundance of other Symbiodinium in the environmental population pool. Laboratory experiments were in accordance with this field observation: Symbiodinium strains of type A1 and D1-4 showed higher infection rates for Acropora larvae than other genotype strains, even when supplied at lower cell densities. Subsequent attraction tests revealed that three Symbiodinium strains were attracted toward Acropora larvae, and within them, only A1 and D1-4 strains were acquired by the larvae. Another three strains did not intrinsically approach to the larvae. These findings suggest the initial establishment of corals-Symbiodinium symbiosis is not random, and the infection mechanism appeared to comprise two steps: initial attraction step and subsequent selective uptake by the coral.