To establish objective and scientific criteria for describing quantitative traits， we evaluated 85 Dendrobium varieties based on 20 quantitative traits， followed by statistical analysis using the least significant difference method and the range method to determine the grading range. The genetic diversity was assessed using the coefficient of variation， genetic diversity index， and correlation， principal component and clustering analysis. The results of K-S normality test showed that 15 quantitative traits， including plant size， pseudobulb length， pseudobulb diameter， blade length， blade width， pedicel length， pedicel diameter， flower longitudinal diameter， flower transverse diameter， middle sepal length， lateral sepal length， lateral sepal width， petal length， petal width， lip petal length， followed a normal distribution and were graded using the least significant difference method. Other traits， including number of flowers， peduncle length， peduncle diameter， middle sepal width， and lip petal width， were graded using the range method. The performance of each quantitative trait was stable within the varieties， and the variation among the varieties was rich， with the coefficients of variation of quantitative traits within and among varieties ranging from 5.44%-17.33% and 24.99%-124.03%， respectively. The genetic diversity indices of the 20 quantitative traits were between 1.16-2.05， indicating the suitability of this collection for quantitative traits grading studies. Correlation analysis revealed generally highly significant or significant correlation among these traits， with large absolute correlation coefficient values. Especially， the correlation coefficients between flower longitudinal diameter and flower transverse diameter， lateral sepal length， petal length， middle sepal length and lateral sepal length， petal length， lateral sepal length and petal length are all above 0.95， indicating a high degree of correlation. Such highly correlated traits could be reduced in the DUS testing process to improve efficiency. Principal component analysis showed the first two principal components accounting for 84.78% of the cumulative contribution， with the first principal component （73.54%） representing flower size， and the second （11.24%） representing the number of flower and plant height. The clustering analysis results showed that the 85 Dendrobium varieties could be divided into two groups at the Euclidean distance of approximately 60， which mainly reflected the differences in plant size， flower longitudinal diameters and flower transverse diameters. Moreover， the second group could be divided into two subgroups at the Euclidean distance of approximately 30， which mainly showed the differences in the number of flower and peduncle length. Combination of principal component and cluster analysis， the number of flower has a strong distinguishing ability among varieties and can be added as a grouping trait for DUS testing of Dendrobium.