Objective: The current study examined whether cognitive reactivity, cognitive extremity reactivity, and mood reactivity following mood provocation predicted relapse in depression over 5.5 years. Additionally, this study was the 1st to examine whether changes in cognitive reactivity and mood reactivity following preventive cognitive therapy (PCT) mediated the preventive effect of PCT on relapse. Method: One hundred eighty-seven remitted recurrently depressed outpatients were randomized over treatment as usual (TAU) versus TAU + PCT with 5.5-year follow-up. Relapse in depression was assessed with the Structured Clinical Interview for "DSM-IV" Axis I Disorders (Spitzer, Williams, Gibbon, & First, 1990). Results: Mood reactivity predicted time to relapse over 5.5 years. We found no evidence that cognitive reactivity was a risk factor for relapse in depression. Moreover, unprimed dysfunctional beliefs predicted relapse directly. There was no indication of mediation by changes in cognitive reactivity (including extremity of the beliefs and unprimed beliefs) or mood reactivity on the preventive effect of PCT. Further, explorative analyses revealed that increases in cognitive and mood reactivity over time also predicted time to relapse. Conclusions: Our findings highlight a need to focus on mood reactivity instead of beliefs as a risk factor for relapse in depression. Similar to a previous study, we found no indications that cognitive therapy after remission reduced dysfunctional beliefs, cognitive reactivity, or extremity. Future studies should examine cognitive reactivity and mood reactivity in daily life as predictors of relapse. (Contains 2 tables and 1 figure.)