Emotional complexity (EC) involves the ability to distinguish between distinct emotions (differentiation) and the experience of a large range of emotions (diversity). Lower EC has been related to psychopathology in cross-sectional studies. This study aimed to investigate (a) whether EC prospectively predicts psychopathology and (b) whether this effect is contingent on stressful life events. To further explore EC, we compared the effects of differentiation and diversity. Adolescents from the general population (N = 401) rated 8 negatively valenced emotions 10 times a day for 6 consecutive days. Further, they completed the Symptom Checklist-90 (baseline and 1-year follow-up) and a questionnaire on past year's life events at follow-up. Logistic regression analyses tested whether EC-reflected by emotion differentiation (intraclass correlation coefficient [ICC]) and diversity (diversity index [DI])-predicted prognosis (good: remitting or lacking symptoms vs. bad: worsening or persisting symptoms). EC predicted prognoses but only when based on the ICC (OREC.ICC = 1.42, p = .02). An ECICC 1 SD above average increased the probability of good prognosis from .67 to .74. This effect was not related to stressful life events (OREC × Life events = 1.03, p = .86) and disappeared when emotion intensity (mean level) was taken into account (OREC = 1.20, p = .20). Predicting future prognosis does not necessitate complex measures of emotional experience (ICC, DI) but rather might be achieved through simpler indices (mean). The discrepant effects of the ICC and DI on prognosis suggest that impaired emotion representation (ICC) plays a more important role in vulnerability to mental ill health than does low diversity of emotions (DI).