Cultivation of safflower ( Carthamus tinctorius L.) especially in summer has a crucial importance in arid and semi-arid region in Iran. This is because of safflower high genetic diversity in this area, high essential unsaturated Linoleic fatty acid (78%), edible oil quality comparative to olive oil, medical usage, drought tolerance, and high adoption. In other hand elements such as wilting and dumping off disease and safflower fly have the huge impact on the safflower yield reduction in the world. This make biotic stress tolerance the most important breeding goal in safflower. The lack of tolerant genes among the cultivated genotpes and the close relation of wild species ( Carthamus oxyacanthus ) with cultivated safflower makes gene integration the appropriate path to transfer desire genes such as biotic stress tolerance. Although Linkage Drag Effect is a restriction of such crosses, nowadays the use of molecular markers to select the background genome of the recurrent parent could help resolve this problem. In this research two backcross populations (BC 1 F 1 ) with more than 250 plants in each population was developed from interspecific hybridization between safflower and its wild relative ( C. oxyacanthus ) during 18 months of several continuous cultivations. One of thepopulation which was the resulte of hybridization between the domesticated parent, C 111 , an Iranian cultivar and the wild parent, Isf1, one of the most important annual weeds in the west of Iran was used for further investigation. In this population 196 plants and the two parental genotypes were evaluated using 23 agronomic traits, 21 SRAP molecular markers and 6 phenotypic markers to determine the efficiency of Marker Assisted Background selection (MAB) versus phenotypic selection. Investigations using agronomic traits indicated that there was an spectrum of individuals with diversity resembling both parental plants. In addition 5 significant PC calculated by principal component analysis showed complex relations between traits which could entirely separate wild parent from the rest of population. Obviously this situation could be caused by the nature of agronomic traits which diffrentiats the wild genotype from cultivated one efficiently. Also cluster analysis could separate this population into several groups with different distance relatively closer to domesticated parent. The SRAP evaluations conducted by 8 primer compositions with maximum PIC for M2/E6 (41%) and minimum PIC for M2/E5 (31%) showed linkage between markers. This linkage did not prevent these markers to differentiate between parental groups resulted by cluster Analysis. In other hand phenotypic markers were not as usefule because of high linkage between markers which was revealed by the Principal Component Analysis and Factor Analysis results. Although SRAP markers could not differentiate the population in the same way as agronomic traits did, the relationships between these two kinds of characteristics can not be disregarded. Some of the most important traits which showed significant relation with most SRAP markers were fertile capitula per total capitula ratio, capitula diameter, capitula height, diameter per height ratio of capitula and capitula weight. In other hand traits which showed significant relations with most phenotypic markers were last fertile hand height and harvest index. Stepwise regression was conducted to indicate the most important yield components involving in plant yield in this research. Results showed a model (95.3%) consist of number of fertile capitula (90.5%), number of seeds per capitula (3.1%), capitula height (1.5%), diameter per height ratio of capitula (0.1%) and capitula diameter (0.1%). Also Path Analysis showed that plant yield were affected by number of fertile capitula first and capitula diameter second. Key words: Carthamus tinctorius , Carthamus oxyacanthus , Inter-specific hybridization, Backcross, MAB, SRAP