Narrative self-evaluation patterns were studied in relation to longitudinal measures of adaptation to the death of a spouse in midlife. Narrative self-evaluations, identified in open-ended interview transcripts at 6 months post-loss, were coded as either positive or negative and as either doing-based (evaluations of "what one does") or being-based (evaluations of "what one is"). These narrative variables were then compared with separate, clinical-interview measures of grief at 6, 14, and 25 months post-loss. Results confirmed 3 predictions. First, participants who made an optimal proportion of positive to negative self-evaluations (approximately a 5:1 positive-to-negative ratio) had lower grief levels over time than did those who made either higher or lower proportions. Second, the tendency to focus on evaluations of what one does rather than what one is predicted lower grief levels over time. Third, participants who directly integrated doing-based and being-based self-evaluations had lower grief levels over time than those who did not link the 2 evaluations. Implications For the narrative construction of personal meaning and identity in relation to adaptation are discussed.