Abstract: The characters of 72 rose cultivars were observed and measured in the International Rose Test Station (Shenzhen Station), and 28 preliminarily chosen characters were studied. Principal Component Analysis (PCA) and Rtype cluster analysis were conducted after the 28 characters were quantitatively coded, and Qtype cluster analysis on the elected principle characters was consequently carried out by SPSS software. It was concluded on the basis of PCA analysis that the rose classification was largely dominated by characters relating to plant growth properties, branching angle, petal and thorn status, etc.. It was showed by the Rtype cluster analysis result that the 28 characters might be divided into several groups, and there were significant correlations among the characters of plant thorn morphology, floral organs and leaf features inside the groups. The 72 rose cultivars were classified into 9 types according to Qtype cluster analysis.