Seventy-nine primitive cotton (Gossypium hirsutum L.) accessions were each crossed to 'Deltapine 16', and progeny with day-neutral flowering habit were selected. These progeny were then backcrossed four times to their respective primitive parents and selected for day-neutrality in the F2 generation following each backcross. The resulting BC4F4 progenies were evaluated for yielding, fiber and other agronomic traits in 1991 and 1992. An additive-dominance genetic model was used to evaluate these traits. MINQUE method was used for estimating genetic variance components and predicting genetic effects. The analyses results indicated that the dominance variance components were not significant for most traits estimated, with the exception of boll size. The additive variance components had relative low values to contribute to the phenotypic variances (VA/VP about 4 - 21%), but they were significant for all traits estimated. Heritabilities (narrow sense) had an average of 17% for these genetic backgrounds; however, boll size had the highest broad-sense heritability (48.8%). The predicted additive effects were mostly positive, but the proportion of significant-positive additive effects varied with germplasm accessions and characteristics, with a range from 6% to 27%. Positive additive correlations were estimated for yield with lint percent; lint percent with boll size and elongation; boll size with seed index; strength with 50% span length, also with first fruiting branch, main stem nodes and plant height; 2.5% span length with 50% span length; and among 4 plant mapping characters. However, all yielding characters were negatively correlated with plant mapping characters and strength. Comparison of the mean performances of these accession progenies with commercial cultivars over two years for 13 characteristics showed that except for lint yield and lint percent, most were equal to or higher than that of the commercial cultivars with extensive variation.