The breeding materials used in this study were 40 families of flax derived from four crosses {P₁ (Giza7) x P₄ (S.402/3/3/10), P₂ (Giza 8) x P₆ (Ariane), P₃ (S.329/2/23/6) x P₅ (S.421/43/14/10) and P₄ x P₆} as well as a bulk of each cross in F₃ and F₄ generations in addition two check varieties (Sakha 1 and Sakha 2). These genotypes were grown in Randomized Complete Block Design with three replicates at Etay El-Baroud Exp. Sta., El-Beheira Governorate during the two successive seasons (2004/05 and 2005/06). The present study was aimed to compare the improvement resulting from application of independent culling levels selection (ICL) method with the hybrid bulk for straw and seed weight in early segregating generations of flax hybrids. The results obtained could be summarized as follows: 1-     Using of ICL method in most cases, was more efficient in improving straw weight per plant through selection for its two important components, plant height and technical length than bulk population. Also, number of capsules per plant and 1000-seed weight could be used as selection criteria to improve seed weight per plant in both of F₃ and F₄ generations . In the meantime these traits gave low discrepancy between phenotypic coefficient of variability (PCV) and genotypic coefficient of variability (GCV) values with high heritability (H) as well as high genetic advance (GA%). 2-     Also, clear wide variation was noticed between mean performances for most studied traits in F₃ and F₄ generations of the four crosses when using ICL selection method compared with other entries (the four bulk crosses and two check varieties) under study. These results indicated the amount of improvement which occurred by using this method of selection. 3-     Selection for straw weight per plant and its two important components in two crosses (P₁xP₄ and P₃xP₅) as well as selection for seed weight per plant and its components in the cross (P₂xP₆) may be recommended for isolating superior genotypes characterized by high straw and seed yields in latter generations. High genetic advance with high heritability may be attributed to a high degree of additive gene effects, for these characters, hence these crosses are likely to respond to direct selection 4-     Phenotypic correlation coefficients among eight characters indicated that maximization of straw weight may be obtained by selection for number of basal branches per plant, plant height and technical length. Seed weight exhibited significant positive correlation with number of capsules per plant and straw weight for all crosses. The positive correlation between straw weight and seed weight per plant, supports the evidence for the possibility of selection genotypes characterized by high straw yielding ability and simultaneous high seed potentialities (dual purpose type).