This research tested hypotheses that negative evaluations of important roles predict vulnerability to stress-related increases in psychological distress, that stressful events are related to increases in distress in part by increasing negative role evaluations, and that domain-specific measures are superior to aggregate measures in making such predictions. In a prospective study controlling for social desirability and initial symptoms, negative role evaluations were associated with increased vulnerability to dysphoric, but not anxious, reactions to negative life events. Negative events were associated with increases in negative role evaluations over time, and these increases accounted for much of the stress-symptom relation. Aggregate measures were better predictors of stress vulnerability, but domain-specific measures had an advantage in analyses of cognitive mediators of life events. Thus, the advantages of using aggregate vs. domain-specific measures may depend upon the aspect of the stress process under investigation.