Evidence supporting the effectiveness of cognitive behavior therapy (CBT) for stress-related illness is growing, but little is known about its mechanisms of change. The aim of this study was to investigate potential mediators of CBT for severe stress in form of clinical burnout, using an active psychological treatment as comparator. We used linear mixed models to analyze data from patients (N = 82) with clinical burnout who received either CBT or another psychological treatment in a randomized controlled trial. Potential mediators (i.e., sleep quality, behavioral activation, perceived competence, and therapeutic alliance) and outcome (i.e., symptoms of burnout) were assessed weekly during treatment. The results showed that the positive treatment effects on symptoms of burnout favoring CBT (estimated between-group d = 0.93) were mediated by improvements in sleep quality, ab = -0.017,95% CIasymmetric [-0.037, -0.002], and increase in perceived competence, ab = -0.037, 95% CIasymmetric [-0.070, -0.010]. Behavioral activation, ab = -0.004 [-0.016, 0.007], and therapeutic alliance, ab = 0.002 [-0.006, 0.011], did not significantly mediate the difference in effects between the treatments. Improving sleep quality and increasing perceived competence may thus constitute important process goals in order to attain symptom reduction in CBT for clinical burnout.