Inadequate knowledge of the population structure and diversity present often hamper the efficient use of germplasm collections. Using a high through-put system, twelve microsatellite loci were used to analyze genetic diversity and population structure in a national field genebank repository of 243 cacao accessions grouped into 11 populations based on their known sources. Based on multi-locus profiles, the Bayesian method was used for individual assignment to verify membership in each population, determine mislabeling and ancestry of some important accessions used in breeding program. A total of 218 alleles was revealed with a mean number of 18.2 alleles per locus. Gene diversity (He = 0.70) and allelic richness (4.34 alleles per locus) were highest in the F1 hybrid population. Differential mating system was suggested as responsible for the observed deficit and excess of heterozygotes observed among the populations. Analysis of molecular variance showed that within-population variance accounted for 63.0% of the total variance while the rest 37% was accounted for by the among-population variance. Cluster dendrogram based on UPGMA revealed two main subsets. The first group was made up of the Amelonado/Trinitario ancestry and the other of Nanay/Parinari ancestry. We found that Nanay and Parinari populations were the major source of Upper Amazon genes utilized while a large proportion of genetic diversity in the field genebank remained under-utilized in development of improved cultivars released to farmers in Nigeria. This study showed that the presence of alleles of the Upper Amazon Forasteros (Nanay, Parinari and Iquitos Mixed Calabacillo) genetic materials in the locally available accessions predated the formal large scale introduction of Upper Amazon materials in 1944. This is the first report of population structure of field genebank collections of cacao in Nigeria since more than seven decades of formal cacao breeding research.