Abstract Twelve microsatellite markers were used to characterize 130 clonal cultures of Perkinsus olseni derived from 30 clams from six different geographic locations. Only two loci were polymorphic in the four populations studied from Spanish coast (mean sample size = 31.2), and a third locus was variable in only two populations. In contrast, five parasites isolated from five clams from Japan and New Zealand showed variation at nine loci. Low genetic variation (2.08 ± 0.64 alleles per locus; mean genetic diversity: 0.101 ± 0.022), and very high F IS values (0.857 on average) were observed in Spanish populations. A total of 39 multilocus genotypes (MLGs) were identified in the whole sample (121 clonal isolates after excluding incomplete MLGs due to missing data). A three-level hierarchical analysis of molecular variance found significant levels of genetic variation within infrapopulations (all the parasites in a single host; Φ IS = 0.679) and among infrapopulations within the component population (all the parasites among a host population; Φ SC = 0.579). Differences among the component population from different geographic locations were not significant ( Φ CT = 0.057). These results suggest that an important fraction of F IS is explained by the Wahlund effect, but also strong inbreeding within infrapopulations. Another explanation for the high F IS within infrapopulations is the presence of haploid and diploid stages in the clam. Although fully aquatic system provides many opportunities for mixing of parasites from different clams, results are consistent with the consideration of all P. olseni in a clam as a cohesive genetic unit (i.e., deme). If the parasite was introduced into the Spanish coast with the importation of infected clams from Asia and Oceania, the low microsatellite polymorphism could be reflecting founder effects in the recent evolutionary history of P. olseni. The loss of alleles would be intensified in a scenario structured in numerous demes because of recurrent founder effects at microgeographic level.