In order to breed the Luffa varieties, with 32 different luffa germplasm resources as research materials, 22 characteristics including bearing fruit habit, leaf margin, fruit ribbing, percentage of nodes with female flowers to total nodes and obtained various data of characteristics were performed using the principal component analysis and the cluster analysis method by the DPS software. The results showed that the 22 characteristics coule be integrated into 5 principal components and the cumulative contribution rate reached 81.308%. According to the correlation between the former 5 principal components and characters, 14 influential characteristics were selected. Based on the principal components, the 32 luffa germplasm resources could be divided into 2 groups by the Euclidean distance on the first level and further divided into 6 subgroups using cluster analysis method.