Abstract: Classfication，genentic diversity of forty rose natural hybrids were studied with the two methods：Numerical taxonomy and SCoT molecular marker. Forty rose natural hybrids were divided into 5 groups by numerical taxonomy. The first group included Z1 and Z15 in all total of 20 samples，the main characteristics were that the color of petals were pink to deep pink，multiplicate flower，coil of calyces were no，and the stalk thoms were even；The second group included Z10 and Z17 in all total of 6 samples，the main characteristics were that the color of petals were purple，single flower，coil of calyces were yes，and the stalk thoms were even；The third group included Z12 and Z38 in all total of 12 samples，the main characteristics were that the color of petals were light pink to pink，multiplicate flower，coil of calyces were no，and the stalk thoms were underside；Z11 was the fourth group and Z34 was the fifth，the main different charactersitic was the color of petals，yellow and white respectively. A total of 426 bands were generated from 20 SCoT primers that had been selected by SCoT，of which 379 bands hadpolymorphism. The percentage of polymorphic bands was 88.97%. UPMGA was selected for cluster analysising in SCoT molecular marker，40 rose natural hybrids also had been divided into 5 groups. The results of numerical taxonomy and SCoT cluster analysis were similar. The two selected methods，numerical taxonomy and SCoT molecular marker were fit for this study. The classification result of 40 rose natural hybrids was scientific and reliable. Forty rose natural hybrids had abundant genetic diversity. Diversity of the parental origin made the rose classification system increase 2 groups compared with the previous results. It could provide a theoretical basis for the scientific utilization of rose natural hybrids and for cultivating new rose varieties.

Key words: Rosa rugosa, natural hybrid, numerical taxonomy, SCoT molecular marker, cluster analysis