Recent longitudinal data suggest that retrospectively defined earned-secures are not more likely than continuous-secures to have been anxiously attached to their mothers in infancy and indeed experience high-quality maternal parenting in childhood. Such findings leave unanswered the question of why earned-secures report negative childhood experiences. On the basis of speculation that earned-security may result from depression-related biases in the recall of early experiences, this report describes the effects of an experimental mood induction on the valence of young adults' (18-25 years) life narratives as assessed in the Adult Attachment Interview. Among secure adults, individuals in a sadness condition were more likely to be classified as earned-secure; happy participants were more likely to be classified as continuous-secure. Induced mood was unrelated to security versus insecurity.