To determine the best distance in honeycomb selection method and investigate response ofcotton growth and seed cotton yield that can be attain it. Field trails were undertaken on the farm of Dept. of Field Crops, College of Agriculture, University of Baghdad during 2009-2010. Cocker130walt and Lashata cultivars were grown. Three distances (130, 105 and 80cm) among plants were applied in honeycomb design. The produced seeds from selection by honeycomb were planted to evaluate the genotypes to compare with original populations, according to a factorial experiment with RCBD under 53 and 89 thousand plant/ha. The results showed superiority of selected Cocker310walt in distance 130x113cm compared with origin. It increased from 107.33 to 136.34cm. The same selected and Lashata at the same distance gave higher no. of sympodia (23.67 and 21.17 branches, respectively), higher no. of total bolls/plant (54.67 and 46.33 boll, respectively), but they gave lower no. of nodes to first sympodial branch and no. of monopodia. The values of C.V, SE and SD decreased in all selected plants that mean increasing homogeneity. Highly heritability narrow sense was obtained from no. of total bolls/plant that associated with higher genetic gain (72.70% and 33.42%, respectively). It was clear that effect of additive gene action for improving this trait. It can be concluded that using of selection was effective to improve selected plants according on studied characters. We recommend to use honeycomb in breeding programs to improve cotton performance.