By extensive investigation in Gansu Province of northwestern China, where flare tree peonies are concentratedly cultivated, this study selects one population that is composed of 150 seedlings to evaluate the adaptability for association analysis in molecular assisted selection. Totally 32 characters were investigated and analyzed, on which phenotypic variation analysis, principal component analysis (PCA) and correlation analysis were conducted. The results showed that the variation of all 22 numerical traits was extremely significant (P0.01) and the coefficient of variation (CV) was 10%-30%, all 10 taxonomic characters had a rich variation in phenotypes, representing phenotypic variations of P. rockii hybrids. PCA showed that the traits of stems and buds and the length of compound leaves were the best predictor of polymorphism patterns in flare tree peonies, and the variation of phenotypes was with equal distribution within this population without conspicuous structure, but a hierarchical relationship was observed among different traits. The above results indicate that this selected population is well suited for the association analysis of charactermolecular marker in P. rockii hybrids.