In order to study the variation of quantitative traits and their relationships with seed yield, an experiment was conducted in Research Field of Agricultural College in 1998, using split plot design in a randomized complete block design with three replications. Main plots were 3 irrigation regimes (non stress, middle stress and terminal stress) and subplots were 8 commercially late-maturing single cross hybrid cultivars. Analysis of variance and mean comparison between treatments showed that the hybrids for all of the traits had considerable variation, and 3 irrigation regimes showed significant differences in several traits. In general, most of the traits related to stress condition had negative response and the maximum injury due to stress was on seed yield. This effect was due to severe reduction on number of seeds
per row, ear lenght and 500 kernel-weight at the flowering and seed filling stages. The results of simple correlations, stepwise regression and path analysis indicated that the selection must be performed on the number of seeds per row and ear diameter in the three irrigated regimes