The experiments were conducted in the farms of Fac. Agric. Assiut Univ. in clay and new reclaimed soils. The plant materials consisted of 41 sorghum entries including 30 F1 hybrids, five restorer lines and six B-lines (cytoplasmic male sterility lines). The testers were crossed to lines in season 2013. The 41 entries were evaluated in 2014 and 2015 seasons under well-watered and drought stress at two locations represent clay and reclaimed soils (eight environments). Surface and drip irrigation systems were applied in the clay and reclaimed soil, respectively in order to carry out the well-watered and drought stress conditions. The combined analysis of variance indicated significant (p≤0.01) differences among the 8 environments, and most of the variability was for environments. Mean squares of genotypes, lines, testers and crosses, and their interaction with environment were significant (p≤0.01) for all traits reflecting sufficient genetic variability, the presence of both additive and non-additive, and assessment of combining ability could be proceeded. The standard deviation of the additive variance under the eight environments was less than the dominance variance for all traits. This indicates that the additive variance was more stable than the dominance one with differences in environment. Except few cases, results indicated that both additive and dominance variances were larger under well-watered than under drought stress, and under clay soil than under reclaimed soil for all traits. The dominance variance was larger than the additive one in 58 of 64 cases of all traits under study. The combined data indicated that the additive variance across environments was larger than the dominance variance only for plant height, however, the dominance variance was larger than the additive one for days to blooming, panicle weight, panicle length, panicle width, grain yield /plant, threshing % and seed index. There was no line or tester could be considered the best combiner for all traits, however line no.3 was the best combiner for most traits. The combinations; F3×T3, F3×T4 and F1×T4 could be considered the best in grain yield / plant and showed significant better parent heterosis.