To identify high-quality safflower germplasm and provide theoretical reference for breeding, 482 safflower germplasm were analyzed on their genetic diversity, correlation, principal component by 12 agronomic traits from 30 counties and regions worldwide. The 12 traits including plant height, branch height, height of the top branch, number of primary branches, number of secondary branches, seedling survival rate, initial flowering, flower color, with or without spines on leaf margin, with or without spines on bract, leaf margin, and thousand-grain weight were selected as agronomic traits for comprehensive evaluation of safflower germplasm. Results showed that abundant genetic diversity was found among these recourses. The genetic variations were mainly from secondary branches number (74.2%), seedling survival rate (60.5%), number of primary branch (52.9%), branch height (51.0%), height of the top branch (27.0%), plant height (23.7%), and thousand-grain weight, (21.2%). Correlation analysis found that plant height had significant positive correlations with other traits except for leaf margin, thousand-grain weight and flower color. Thousand-grain weight had significant positive correlation with the number of secondary branches, and had negative correlations with plant height, branch height, and height of the top branch, respectively. The bract trait also represented significant correlations with branch height, height of the top branch, and leaf margin traits. Principal component analysis revealed that four components contribute 65.882% of all components. The 1st component included height of the top branch and bract traits. The 2nd component included primary and secondary branches. The 3rd component included leaf and bract traits, and the 4th component included survival rate and flower color. By cluster analysis, 482 safflower germplasm was divided into 5 groups at the genetic distance of 7.5. In summary, it indicated that safflower resources had rich genetic diversity. The 12 agronomic traits above could be used on effectively resources evaluation from different regions.