BACKGROUND: There is increasing evidence that mindfulness-based interventions reduce psychological distress in various medical populations. However, it has hardly been studied if these effects are mediated by an increase in mindfulness. The aim of this study was to examine mediating effects of various mindfulness facets on effects of a Mindfulness Based Cognitive Therapy (MBCT) on perceived stress and mood. METHODS: Outpatients with diabetes types 1 and 2 and low levels of emotional wellbeing were randomized into a group receiving MBCT (n=70) or a waiting-list control group (n=69). Primary outcomes were mood and perceived stress. Before, after and at follow-up (6months post intervention) relevant questionnaires were completed. RESULTS: Mediation analysis using bootstrap resampling indicated that increases in total mindfulness and the facets observing and nonreactivity mediated the effects of the intervention on depressed and angry mood, anxiety (only observing), and perceived stress (only nonreactivity) from pre- to post-intervention. In contrast, from post-intervention to follow-up, besides total mindfulness the facets of acting with awareness and nonjudging mediated the effects on depressed, anxious, and angry mood, while only nonjudging mediated the effects on perceived stress. DISCUSSION: The findings indicate that increases in levels of mindfulness mediate the effects of the Mindfulness-Based Cognitive Therapy in patients with diabetes. It is notable that different facets may be important for immediate change versus long-term outcome. These findings might be relevant for post-intervention care.