Prunus sibirica is an economically important tree species that occurs in arid and semi-arid regions of northern China. For this species, creation of a core collection is critical for future ecological and evolutionary studies, efficient economic utilization, and development and management of the broader collection of its germplasm resources. In this study, we sampled 158 accessions of P. sibirica from Russia and China using 30 pair of simple sequence repeat molecular markers and 30 different schemes to identify candidate core collections. The 30 schemes were based on combinations of two different sampling strategies, three genetic distances, and five different sample sizes of the complete germplasm resource. We determined the optimal core collection from among the 30 results based on maximization of genetic diversity among groups according to Number of observed alleles (Na), Number of effective alleles (Ne), Shannon’s information index (I), Polymorphic information content (PIC), Nei gene diversity (H) and compared to the initial collection of 158 accessions. We found that the optimal core collection resulted from preferred sampling at 25% with Nei & Li genetic distance these ratios of Na, Ne, I, PIC and H to the complete 158 germplasm resources were 73.0%, 113%, 102%, 100% and 103%, respectively, indicating that the core collection comprised a robust representation of genetic diversity in P. sibirica. The proposed core collection will be valuable for future molecular breeding of this species and management of its germplasm resources.