Safflower is a neglected crop with many traditional and industrial uses form high quality edible oil to cheaper substitute for saffron. In this study, 81 safflower genotypes were evaluated in a 9×9 simple lattice design for number of plants per plot (NPP), plant height (PH), height of the first lateral branch (HFL), height of the first lateral capitulum (HFC), stem diameter (SD), number of lateral branches per plant (NLB), number of main branches per plant (NMB), number of capitula per plant (NCP), number of seeds per main capitulum (SMC), number of seeds per lateral capitulum (SLC), seed yield (SY) and thousand seed weight (TSW). The genotype by trait (G×T) interaction biplot model was used to indicate the pattern of G×T interaction two-way interaction data in a graph and showed that the first principal component effect explained 49%, and the second principal component, 24%, of the total mean squares interaction. The vector-view displayed that NCP with NMB, and SMC with SY were positively associated while there was a negative association between HFC with TSW, and between NLB with NPP. The polygon-view graph is divided into eleven sectors, and the sector of genotype G80 was winner for most traits. Genotype G58 followed by genotypes G30, G33 and G72, were the most favorable genotypes in the regard of SY while regarding this trait as a reference, SMC was identified as most related trait which is followed by SLC, SD. Applying G×T biplot to the safflower multiple trait data demonstrated that this model graphically displayed the interrelationships among seed yield with number of seeds per main and lateral capitula followed by number of capitula per plant and thousand seed weight, and facilitated visual genotype comparisons and selection. It was found that selection for seed yield alone was not only dependent to number of seeds per main and lateral capitula, but also related to the other traits in safflower breeding.
