Analyses of three waves (6 years) of the National Longitudinal Survey of Adolescent Health data explored the prevalence and stability of sexual orientation and whether these two parameters varied by biologic sex, sexual orientation component (romantic attraction, sexual behavior, sexual identity), and degree of component. Prevalence rates for nonheterosexuality varied between 1 and 15% and depended on biologic sex (higher among females), sexual orientation component (highest for romantic attraction), degree of component (highest if "mostly heterosexual" was included with identity), and the interaction of these (highest for nonheterosexual identity among females). Although kappa statistics testing for temporal stability across waves were significant, they failed to reach acceptable levels of agreement and could be largely attributable to the stability of opposite-sex rather than same-sex attraction and behavior. Migration over time among sexual orientation components was in both directions, from opposite-sex attraction and behavior to same-sex attraction and behavior and vice versa. To assess sexual orientation, investigators should measure multiple components over time or abandon the general notion of sexual orientation and measure only those components relevant for the research question.