Recurrent selection programmes for both high response and low aggression have been employed in the Texas population of Drosophila melanogaster. Five main points have emerged from this investigation. First, the population exhibits extensive genetic variation for the aggression and response components of competitive interactions which take place in genetically heterogeneous cultures. Secondly, the two components of such interactions, namely aggression and response, can be adjusted by the selection of particular groups of genes. Thirdly, the rapid change in the selected components in the early generations suggests that a considerable amount of additive genetic variation is involved in the control of aggression and response. This is further supported by estimates of the realised heritability taken over the whole selection programme which amount to O•79 and O•74 for the low aggression and high response selections respectively. Fourthly, the rapidity with which changes in aggression and response approached plateaux suggests that such changes in the earlier generations are primarily due to the assortment of major chromosomes as units. Fifthiy, it was concluded that aggression and response do not behave entirely independently or dependently. The results suggest that a single array of genes might be responsible for the determination of both characters. The observed results would then be consistent with certain of these genes having a much larger effect on aggression than they do on response.