Although the efficacy of Cognitive Therapy (CT) for depression is well documented, questions remain about the conditions under which CT is maximally effective, and the mechanisms by which CT is efficacious. This study examined the role of extreme responding (ER) as a mediator and predictor of relapse, as well as a predictor of acute treatment outcome. To test mediation, ER of participants in CT was compared to ER of participants receiving Behavioral Activation (BA). Results indicated that ER was not a mediator of relapse in either condition. Both CT and BA showed reductions in depression severity, however neither group showed significant changes in ER over the course of treatment. Finally, ER did not predict or moderate acute treatment outcome. The implications of these results, strengths and limitations of the current study, and future research directions are discussed.