Abstract: In this study, 75 germplasm resources in the primary core collection of Cornus officinalis were collected from three provinces in China. Many quantitative traits, including leaf length, leaf breadth, fruit length, fruit broad, fresh fruit 100-seed weight, seed 1000-seed weight, seed moisture content, ash content of date bark, moisture content of date bark, water soluble extractives and content of Loganin, were analyzed on coefficient of variability and correlation. The core collections were sampled by Euclidean hierarchical clustering method from the primary core collection, and then the construction of the core collection of C. officinalis was evaluated. The results showed that the coefficient of variability in leaf length, leaf breadth, fresh fruit 100-seed weight, seed 1000-seed weight, ash content of date bark and content of Loganin were higher than 10%. It indicated that there were large changes and it was not instability in these characteristics between individuals. The coefficient of variability in fruit length, fruit broad, seed moisture content, moisture content of date bark and water soluble extractives were less than 10%, which indicated that there were less variation and these traits were relatively stable between individual. Correlation analysis results showed that there was a relationship in these traits. The fruit length and fruit breadth had a significant correlation to fresh fruit 100-seed weight (0.396b and 0.312b), which suggested that it was a key reason that the fruit of length and breadth affected the fresh fruit 100-seeds weight. There was significant correlation between the fruit length, the fruit breadth, the content of Loganin and water-soluble extractives, respectively (0.481b, 0.280a and 0.372b). It meaned that the characteristic of water-soluble extractives contents closely related to the fruit size, the fruit shape and content of Loganin. These results will provide a parameter for screening and evaluating the germplasm resources of C. officinalis in the future. There were only 22 samples (29.33% in total) after the core collection was constructed from primary core collection by least distance stepwise sampling (LDSS). The results of T verification and the poor compliance rate (CR%=90.63 > 80%) showed there was no significance between core collection and the primary core collection of C. officinalis. The results showed that the core collection was a better representative sample of the primary collection with minimum repetitiveness and maximum genetic diversity of C. officinalis. The methods of construction and the core collection will provide a reference and research basis for construction of the core collection to C. officinalis or the other medicinal plant resources.