The F₂, F₃, and F₄, generations for ten crosses resulting from a five parent half-diallel were evaluated for yield and fiber traits. Single F₂ and F₃ plants were visually selected for yield from each cross. Ten F₃ and ten F₄ individual plant selections per cross (selected in F₂ and F₃ respectively) and F₃ and F₄ selected bulks for each cross were evaluated for yield and fiber traits in the same test year as the unselected populations. The objective was to quantify heterosis and inbreeding depression and examine progress from early generation selection. The percent heterosis over the high parent for seed cotton yield ranged from -3 to 16% in the F₂, -2 to 19% in the F₃, and -8 to 11% in the F₄ for the ten crosses. The percent inbreeding depression from the F₂ to F₄ generation ranged from -11 to 13% in the unselected bulk populations and -6 to 18% for the F₂ to F₄ selected bulks. There was a much smaller range in percent heterosis for the fiber traits. In general, the yield means of the F₃ and F₄ unselected bulks exceeded those of the F₃ and F₄ selected bulks. This suggests that early generation yields selection may not be effective in increasing the population mean. The expression of heterosis in early generation and subsequent inbreeding depression could hinder progress from early generation selection. However, there were individual plant selections in each cross that exceeded the high parent for yield, 2.5% span length and T, fiber strength, yet there was generally little to no correlation between the percent heterosis or percent inbreeding depression for yield with the number of superior individual plant selections over the ten crosses. There were moderate correlations between the ratio of additive to dominance effects for seed cotton yield with the number of superior F₃ and F₄ single plant selections. This suggests that early generation selection was more effective in crosses where additive effects were of greater importance than dominance effects. The highest correlation (r= .58) was foumd between the E, for seed cotton yield and the number of superior single plant selections over the ten crosses. A practical solution to the need for reducing test material in early generation would be to select superior crosses based on F₂ performance for yield. Also, the preponderance of additive effects for 2.5% span length, elongation and fiber strength suggests that early generation selection could be practiced for these fiber traits in addition to or simultaneous with yield selection among crosses.