With data from a junior high school in an urban area of North Carolina we use log-odds ratios and a nested contingency table model to investigate homogeneity bias in adolescent friendship choice along the dimension of sexual behavior. Females, both black and white, tend to name friends whose sexual behavior is like their own even when grade, other more visible deviant behaviors, and a composite index of deviance proneness are taken into account. In contrast, neither sexual intercourse nor any other deviant behavior considered in this analysis is a significant factor in accounting for male adolescent friendship structure. We suggest that the differential reputational consequences for males and females who have intercourse may account for the observed difference by sex in the friendship choice-sexual behavior relationship.