University students, masked to group status, judged the physical attractiveness of girls with gender identity disorder and clinical and normal control girls, whose photographs were taken at the time of assessment (mean age, 6.6 years). Each student made ratings for all girls for five traits: attractive, beautiful, cute, pretty, and ugly. A multivariate analysis of variance showed a significant group effect. Multiple comparisons of the significant univariate effects showed that the girls with gender identity disorder had significantly less attractive ratings than the normal control girls for the traits attractive, beautiful, and pretty who, in turn, had less attractive ratings than the clinical control girls. Girls with gender identity disorder and the normal controls also had less attractive ratings than the clinical controls for the trait cute. Correlational analyses showed that age was substantially negatively related to the attractiveness ratings in the group of girls with gender identity disorder, but was considerably less so in the two control groups. The extent to which the group differences in attractiveness were due to objective, structural differences in facial attractiveness vs. socially created, or subjective, processes is discussed.