- Five winter wheat cultivars and their diallel crosses were evaluated for plant height, harvest index, deading-maturity duration, the components of yield, (spikes per plant, spikelets per spike, kernel weight and kernels per spikelet)and total plant yield. Two diverse locations, Moro, a dryland site (250 mm annually) located in central Oregon and the Hyslop Agronomy Farm, a high rainfall site (over 1000 mm annually) located in the Willamette Valley, were utilized for one and two cropping seasons, respectively. Three rates of seeding were used as main plots in a split-plot design that was replicated four times. A modified blend method of seeding was used to simulate solid seeding conditions. Experimental seeds were planted 30.5 centimeters apart within the row over a filler cultivar in equally spaced (30.5 centimeters) rows. The data were analyzed by analysis of variance, Griffing's diallel analysis (Method 4, Model 1), correlation, path-coefficient analysis and by parent progeny regression. The correlation between grain yield, its components, harvest index, maturity-duration and plant height was dependent on the particula. r environment of the test. The re waspoorcorrelationbetween yield, tiller number and seed size under all the conditions of these studies. Negative associations between the components of yield indicated the sequential compensatory behavior of these characters under all environments. It would be very hard to select for large grain and short stature wheat because of the positive correlation between plant height and seed size within this population. The low correlations of yield with tiller number and seed size were mainly caused by indirect negative effects through one or more of the other yield components. Harvest index, maturity-duration and plant height had very small direct or indirect influences on grain yield. It was concluded that maximum yield would be obtained from a plant type which produces enough tillers to cover a particular unit of field area with large, fertile spikes, having medium to large kernels and semi-dwarf stature. No significant differences existed between parents and single crosses in the expression of the yield components. Nevertheless, hybrids outyielded their parents in grain yield and demonstrated that heterosis for complex traits was a consequence of multiplicative relationships among the components of these traits. Significant interactions between the genotypes and locations, seeding rates and years were observed in the expression of all characters studied. These interactions indicated that using data from non-competitive conditions to assess performance under competitive conditions could not be justified. Also, limiting the number of testing sites may lead to unsound generalizations and erroneous recommendations regarding gene action estimates of yield and the components of grain yield and three associated characters. Under non-competitive conditions, estimates of the additive type of gene action were more significant and contributed larger effects than the non-additive type for all traits. As competition increased at higher seeding rates, the effects of specific combining ability became more important in the expression of yield, number of spikes, spikelets per spike, and plant height. Heritability estimates confirmed these results except for yield. Of the agronomic characters, harvest index, maturity-duration and plant height, only harvest index showed some promise as a selection criterion under noncompetitive conditions. A breeding procedure utilizing the component approach consisted of selecting early generations under spaced planting with emphasis on avoiding extreme values in any of the components of yield. The balanced combinations of the components of yield should be tested under solid seeding conditions.