The disidentification hypothesis predicts that African American boys achieve less in school than African American girls do because boys have less personal investment in doing well academically (i.e., they are disidentified). When do such gender differences emerge? Using self-perception and achievement data from longitudinal studies of children (N = 113) at high risk for academic problems because they come from low-income families, the authors examined whether elementary school-aged and early adolescent African American boys are more prone to low achievement and disidentification than African American girls. Multiple regression analyses indicated no gender differences in reading or mathematics achievement between boys and girls at age 8 or at age 12. At age 12, African American boys' self-esteem was predicted by academic performance in ways similar to that of African American girls. Thus, no gender differences emerged in elementary school achievement and no gender-specific disengagement patterns were confirmed among at-risk African American students. (Contains 2 notes and 5 tables.)