Dysgenic and non-dysgenic base populations were made by reciprocal crossing of Harwich (P) and Canton-S (M) strains. From each cross, two up and two down selection lines were established, with selection on abdominal bristle number for ten generations. The intensity of selection was 10 out of 50 individuals from each sex. Mean bristle number, phenotypic variation and heritabilities were compared between dysgenic and non-dysgenic populations under selection. Except for an anomalous non-dysgenic downline in which a mutation of large effect occurred, all lines showed similar responses to selection. These results contrast with the results reported by Mackay (1984, 1985) in which substantial increases were obtained for response to selection, phenotypic variation and heritability in the dysgenic compared to non-dysgenic lines. There are some indications that the higher response in our aberrant non-dysgenic downline is the result of transposition. Possible explanations for the occurrence of transposition and dysgenesis in the lines derived from nondysgenic crosses are discussed.