Some selection procedures i.e. selection index involving 11 indices and direct selection for four separately traits (lint yield/plant, bolls/plant, seeds/boll and lint/seed) were used to improve lint yield, yield components and fiber properties in early segregating generations; F₂, F₃ and F₄ of the cotton cross;(Giza 85 x Dandara). Means of F₄ generation were higher than those ofF₂and F₃ generations for all studied characters except for fiber fineness (desirable)and fiber length .The heritability estimates obtained in F₂,F₃ and F₄ generations were ranged from moderate to high (56.18 to 92.2%) for most traits. These estimates indicate a possible success in the selection of the early generations that were evaluated . Estimates of PCV and GCV in F₄ for most traits were higher than those of F₂  and F₃ generations. It was due to the application of several procedures that having various performance in selection. Phenotypic variances (V_p) were found to be greater than the corresponding genotypic variances (V_g) for all studied characters indicating that the expressions of these characters were influenced by the environmental factors. Deviations of the realized advance from the predicted of lint yield determined in F₃ to F₄ generations were positive and high for most selection procedures. Data showed that indices; I._w1, I._w2 , I._w123 ,I._w12 , I._w23 , I._w13, I._w3 and I._xw gives high value of improvement in lint yield as predicted and actual advance   . However, The data also indicated that predicted advance from F₃ had high value of gains relative to actual advance for most indices . Also, the gains were higher in F₃ and F₄ than those F₂ generation. This may be attributed to the efficiency of selection procedures application in this study. After two cycles of selection, the genetic gains from selection isolated it isolation best ten families on base of the highest value in selected and unselected traits; it my be very important in cotton improvement programs  .