Study of genetic variation for agronomic and morphological traits is essential to select appropriate genotypes and increase seed yield in safflower. An experiment was conducted at the Research Farm of Isfahan University of Technology to study genetic variation of agronomic traits and seed yield components among some safflower F2 populations and their parents in 2015. In this study, 9 populations of F2 generation along their parental lines (one foreign and 6 local genotypes) were evaluated in a randomized complete block design with three replications for traits of days to 50% emergence, days to flowering, days to 50% flowering, days to maturity, plant height, seed yield components, seed yield, and seed oil content. The analysis of variance results showed that there was genetic variation among the populations and the parental lines for most of the studied traits. The highest and lowest seed yield per plot among the parental lines belonged to IUTKh211 (with 2642 kg/ha) and IUTH13 (with 957 kg/ha), respectively. Among the F2 populations, the cross IUTc129 × IUTE1449 with 2364 kg/ha and cross Saffire × IUTH13 with 1122 kg/ha had the highest and lowest mean of seed yield, respectively. In terms of oil content, that is an important economical trait in safflower, the lowest and highest mean of this trait was observed in cross IUTH13 × IUTM12 (with 22.9%) and IUTKh211 × IUTH13(with 26.2%), respectively. Among the parental lines, foreign cultivar of Saffire with 26.4% and local genotype IUTM12 with 22.13% had the highest and lowest seed oil content, respectively. A relatively high broad-sense heritability and a low difference between genetic and phenotypic coefficients of variation were observed for most of the studied traits. The highest broad-sense heritability was estimated for days to flowering (93%) and the lowest for heads per plant (24%). Genetic and phenotypic correlation between seed yield components showed that heads per plant and seeds per head had positive and significant correlation with seed yield. The result of stepwise regression also showed that heads per plant, seeds per head and 100- grain weight explained 79% of the variation for seed yield per plot. The results of path analysis for seed yield per plot showed that heads per plant and seeds per head and head diameter in order of importance had more direct and positive effect on this trait. These results showed that heads per plant was the most important component of seed yield per plot, and it can be used as a selection criterion to improve this economical trait in safflower. Cluster analysis separated the parental lines and populations into 3 distinct groups which comprised 3, 9 and 4 population or parental lines and the second group had the highest average of seed yield and its components. Keywords: Safflower, genetic variation, seed yield components, broad-senseheritability, agronomic traits