Forty-five triple test cross progenies produced by crossing 15 homozygous strains of wheat with three testers (varieties HD 2160 and NP 846 and their F1) and 17 parents, were raised in a randomized-block design in two environments (normal and stress). The data on five randomly chosen plants from parents and F8 and ten plants from F1 x lines were recorded for number of days from sowing to heading, plant height, the number of tillers per plant, total biomass, the number of grains per ear, 1000·grain weight and grain yield per plant. The additive and dominance components of genetic variation were each estimated by three methods to determine their relative usefulness in estimating these components. The variance of parents gave better estimates of the additive component than the two other methods (1i + 2i and 1i + 2i + 3i). The method 1i + 2i + 3i proved its superiority over the 1i + 2i method. The dominance component was better estimated by the method 1i - 2i than by the 4σ2d + 9σ2 (1i + 2i - i) and 25 (σ2 (21i - i) pooled with σ2 (22i - i)) methods. The two methods 4σ2d + 9σ2 (1i + 2i - i) and 25 (2σ2 (21i - i) pooled with σ2 (22i - i)) provided similar results for the estimation of dominance component.