Background: The move from inconsistent and problematic autobiographical narrative to a more coherent and reality-based narrative construction of the Self has been discussed as potential mechanism of change in psychotherapies for personality disorders. So far, little empirical evidence exists that demonstrates in a time-dependent design the role of narrative construction in the treatment of borderline personality disorder, in particular when it comes to understanding the integration of body-related information from the affective system with the autobiographical narrative. The present study aims at demonstrating change in emotion-based narrative markers over brief psychiatric treatment and to assess the impact of these changes on subsequent symptom change. Methods: A total of N = 57 clients with borderline personality disorder were assessed at three timepoint over the course of four months of brief psychiatric treatment, within the context of a secondary process-outcome analysis of a randomized controlled trial. Symptom change was assessed using the OQ-45.2 and emotion-narrative change was assessed using the Narrative-Emotion Process Coding System to code client's in-session speech in terms of problem, transition and change markers. Results: All three emotion-based marker categories evidenced significant changes in the assumed direction. The reduction in problem emotion-based narrative markers (e.g., empty story telling) between session 1 and 5 into the treatment predicted the symptom reduction assessed between session 5 and 10. Conclusions: Emotion-based narrative construction may be a suitable method to study the pathway of change toward a more coherent and reality-based narrative construction of the Self-in-interaction-with-the-Other. Reduction of emotion-based problem-marker may be a promising candidate for a mechanism of change in treatments for personality disorders which should be tested in a time-dependent controlled design.