Behavioral inhibition (BI) is a temperamental style characterized by cautious and fearful behaviors in novel situations. The present multi-method, longitudinal study examined whether young children's observed and parent-reported BI in social versus non-social contexts predicts different long-term psychosocial outcomes. Participants (N = 279) were drawn from a longitudinal study of socioemotional development. BI in social contexts (“social BI”) was measured via children's observed wariness toward unfamiliar adults and peers at 24 and 36 months and parents’ reports of children's social fear/shyness at 24, 36, and 48 months. BI in non-social contexts (“non-social BI”) was measured via children's observed fearful responses to masks and novel toys, and parents’ reports of children's distress to non-social novelty at 9 months and non-social fear at 48 months. At 15 years, anxiety was assessed via adolescent- and parent-reports, and global internalizing and externalizing problems were assessed via parent-reports. Confirmatory factor analysis showed that a two-factor model fit the BI data significantly better than a single-factor model, providing evidence for the dissociation of BI in social versus non-social contexts. Social BI was uniquely associated with adolescent social anxiety, whereas non-social BI was specifically associated with adolescent separation anxiety. Neither social BI nor non-social BI predicted global internalizing and externalizing problems, providing evidence for the specific relations between BI and anxiety problems. Together, these results suggest that young children's inhibited responses in social versus non-social situations predict different subtypes of anxiety problems in adolescence, highlighting the multifaceted nature of BI and the divergent trajectories of different anxiety problems.