Objective: Affective associations are key predictors of health-relevant action that can mediate the relationships between cognitive variables and health behavior. Little data, however, has examined affective associations in high-risk groups, the individuals with the greatest need for intervention. Further, few studies have examined the unique predictive ability of positive and negative affective associations. Methods: The present study assessed positive affective associations, negative affective associations, cognitive beliefs, subjective norms, and reports of 4 obesity-related behaviors (physical activity, fruit consumption, vegetable consumption, and soft drink consumption) across a cross-sectional (n = 1,499) and a longitudinal sample (n = 104) of predominantly African American adolescents of low socioeconomic status from the inner city of urban Detroit. Results: In both samples, positive (not negative) affective associations were the strongest and most consistent predictor of health behavior. Analyses also suggested that positive affective associations mediate the relationships between cognitive beliefs, subjective norms, and health behavior. Interaction tests revealed no evidence that the link between positive affective associations and health behavior is moderated by negative affective associations, cognitive beliefs, subjective norms, sex, or age. Conclusions: The results are consistent with affective association research and support the development of health interventions for urban minority youth aimed at changing positive (rather than negative) affective associations. The consistency across behaviors and the lack of reliable interactive effects suggests that changing positive affective associations may benefit urban youth regardless of behavior domain, age, sex, negative affective associations, subjective norm, and cognitive beliefs.