Background: The adjunctive effects of exercise on cortisol levels in people with major depressive disorders (MDD) are equivocal. Therefore, we aimed to compare pharmacotherapy alone (antidepressants) versus an adjunct intervention of pharmacotherapy plus exercise on cortisol levels over 12 weeks. Methods: Serum cortisol were assessed before (baseline-t0), during (Week 5-t1), and at the end of treatment (Week 12-t2). The exercise intervention focused on promoting perceived effort, enjoyment, and overall pleasure throughout the multimodal exercise sessions. The study enrolled 59 adults and randomly allocated them into two groups: exercise group (EG; n= 26,76.9% females, mean age 28.5 years; normal cortisol-EG-CN, n= 18 and hypercortisolemia-EG-HC, n = 7) and control group (CG, n= 29,72.40% females, mean age 26.0 years/normal cortisol CO-CN, n= 17 and hypercortisolemia-CO-HC, n= 11). Results: No changes in cortisol levels were observed over time in both groups (EG and CO). However, in subgroup analysis, participants with hypercortisolemia reduced their cortisol levels regardless of the intervention allocation. At the end of the study, hypercortisolemic MDD participants of EG showed similar cortisol levels with those of the normal cortisol group (EG-CN) [p= .507,d=-.22(-1.11,.67)], while hypercortisolemic MDD participants under exclusive pharmacotherapy (control group) continued to show significantly higher levels than CO-CN [p=<.0001,d=-2.32(-3.41,-1.20)]. Conclusion: Our preliminary findings suggest that, in people with MDD and hypercortisolemia, a structured, multimodal, affect-adjusted, and supervised exercise program along with pharmacotherapy normalized cortisol levels. Notwithstanding, neither pharmacotherapy nor exercise plus pharmacotherapy changed cortisol levels among those with normal cortisol levels at baseline. These results indicate that combining exercise with pharmacotherapy may be effective in reducing cortisol levels specifically in individuals with MDD and elevated cortisol. However, further studies with larger samples are needed to explore this response more thoroughly.