The success of plant breeding operations relies heavily on the nature and extent of genetic components of variation. Thus it is imperative to have reliable estimates of such components in order to formulate an efficient breeding strategy. In the present  study, ten diverse lines in F₂ generation were crossed to three testers viz., Giza 86(L₁) ,Austuralian(L₂) and their F₁ Giza86 X Austuralian (L₃). The parents (lines and testers) and crosses were evaluated in randomized complete block design.  Analysis of variance  revealed significant differences among progenies. The deviations (cultivars L₁ + cultivars L₂ – cultivars 2 L₃) were analyzed to provide a test of epistasis. Additive- dominance model was fitted to the data for these traits not influenced by gene interaction. The results indicated that total epistasis was insignificant for all traits. The partitioning of the total epistasis, showed significance of ( i ) types fixable part of epistasis for all traits except lint  percentage, fiber strength and micronaire, while the unfixable epistasis, dominance x dominance and dominance x additive was significant  for all traits except lint percentage, fiber strength and  micronaire. Additive gene action played the important role for all studied traits except for lint cotton yield, upper half mean, fiber strength and micronaire.  While, dominance effects were insignificant for all traits. The degree of dominance revealed that the ratio was less than unit  indicated the predominant role of additively in controlling boll weight ,seed and lint cotton yield ,seed index and uniformity ratio,  whilst this ratio was equal to unity for lint percentage and upper half mean showed the important of both additive and dominance properties of genes, The higher degree of dominance for fiber strength and micronaire showed over- dominance. The results showed that the dominant alleles were dispersed between testers, as hybrids did not show any proof of directional dominance for all characters.