In 2004 season, twenty one cotton genotypes were selected including seventeen strains along with four commercial cultivars as check varieties were grown in a • randomized complete block design with six replications in six trials (8) at six different locations. Nine traits which included earliness index, seed cotton yield, lint yield, weight of SO bolls, lint percentage, upper half mean, uniformity ratio, micronaire reading and fiber strength were estimated. The studied traits showed highly significant mean squares for genotypes, environment and genotypes and genotypes x environment interaction except micronaire reading which was insignificant (or genotypes and interaction between genotypes x environment. The genotypes varied for the estimate (A) while the estimate (a) did not differ from a = 0 which may suggest that the relatively unpredictable components of the (G x E) interaction variance may be important than other components. It could be concluded that the promising strain F_s 1112/2002 and F6/1193, F₆/1208 and G.84 x (G 74 x G 68) exhibited high yield potentiality and average degrees of phenotypic and genotypic stability. The average phenotypic and genotypic stability was recorded by G.88 for weight of SO bolls traits. While the hybrid G.84 x (G 74 x G 68) exhibited average degree of genotypic and phenotypic stability for lint percentage. For fiber length 2.5%, the two strains F5 1153/2002 and F6 1193/2002 exhibited average degree of phenotypic and genotypic stability. While, the two cultivars G.45 and G.87 exhibited average degree of genotypic and phenotypic stability for pressely index. The results indicated that the best strains were: F₁₁ 1257/2002 which was derived from the cross G.77 x Pima S6 x G.87 x G.77 x G.70) and F12 1292/2002 which was derived from G.87 x G.77 x G.70) which recorded average level of stability for most traits.