It is assumed that chronic or extensive release of cortisol due to stress has deleterious effects on somatic and psychological health, making interventions aiming to reduce and/or normalize cortisol secretion to stress of interest. Cognitive-behavioral stress management (CBSM) has repeatedly been shown to effectively reduce cortisol responses to acute psychosocial stress. However, the effects of CBSM on psychoneuroendocrine responses during "real-life" stress have yet not been examined in healthy subjects. Eight weeks before all subjects took an important academic exam, 28 healthy economics students were randomly assigned to four weekly sessions of cognitive behavioral stress management (CBSM) training or a waiting control condition. Psychological and somatic symptoms were repeatedly assessed throughout the preparation period. Salivary cortisol (cortisol awakening response and short circadian cortisol profile) was repeatedly measured at baseline and on the day of the exam. In addition, cognitive appraisal was assessed on the day of the exam. Subjects in the CBSM group showed significantly lower anxiety and somatic symptom levels throughout the period prior to the exam. On the day of the exam, groups differed in their cortisol awakening stress responses, with significantly attenuated cortisol levels in controls. Short circadian cortisol levels did not differ between groups. Interestingly, groups differed in their associations between cortisol responses before the exam and cognitive stress appraisal, with dissociation in controls but not in the CBSM group. The results show that CBSM reduces psychological and somatic symptoms and influences the ability to show a cortisol response corresponding to subjectively perceived stress. In line with current psychoneuroendocrine models, the inability to mount a cortisol response corresponding to the cognitive appraisal in controls could be a result of a dysregulated HPA axis, probably as a consequence of longlasting stress.