The study investigates the genetic diversity and the ability of genomic-wide selection to predict breeding genomic values of an E. benthamii trial. All individuals (115) of the breeding population were genotyped with 13 microsatellites loci. The diameter at breast height and total height were measured. The data analysis was carried using the softwares: Structure, Popgene, GDA, SPAGeDi1.5 and R. Predictive ability, heritability and standard errors markers were estimated using the RRblup method. The average number of alleles per locus was nine, and the polymorphism level for each locus varied from 3 to 17. The average expected heterozygosity (He=0.655) was very similar to observed heterozygosity and the estimated inbreeding (F = 0.02) was very low. These results corroborate that this population is in Hardy-Weinberg equilibrium for the most loci. The trial genetic diversity is considered high, once the trial sampling demonstrated similar values to the natural populations. The group coancestry (0.085) demonstrate that the trees, in general, related at the half-sib level in this population. By using the Evanno’s method it is inferred that the individuals came from two original populations. The genetic distance calculated among the two groups was low ( =0.21). The heritability estimated from genomic selection for phenotypic traits was very low; however, the heritability estimated using the kinship coefficients was higher. The marker-based heritability using kinship coefficients probably is the more accurate than the one estimated using genomic selection, showing that the population samples can be used to establish breeding populations, hybrids and enriching the species germplasm bank.