In an empirical study, we addressed the temporal properties of personality disorders (PD) and their treatment by schema-focused psychotherapy. We explicitly investigated ‘enduring patterns’, assumed to be the core of PD, and the ‘change mechanisms’ of psychotherapy process. The goal was to introduce a method by which psychotherapeutic change mechanisms of PD patients can be modeled in the temporal domain. 69 patients were assigned to a specific schema-focused behavioral group psychotherapy, 26 to social skills training as a control condition. The largest diagnostic subgroups were narcissistic and borderline PD. Both treatments offered 30 bi-weekly group sessions of 100 min duration each. The statistical approach focused on using time-series panel analysis of the patient-assessed factors Clarification, Bond, Rejection, and Emotion. These factors derived from repeated session-report data obtained after each session. The method provided a detailed quantitative representation of therapy process via the time-lagged associations of these factors. It was found that clarification played a core role in schema-focused psychotherapy, reducing rejection and regulating patients’ emotion. This was also a change mechanism linked to therapy outcome. The introduced process method allowed to highlight the mechanisms by which treatment became effective. Additionally, process models depicted the actual patterns that differentiated specific diagnostic subgroups. Limitations are that time-series analysis explores Granger causality, an approximation of causality based on temporal sequence. In principle, yet unlikely, causal effects of unobserved variables may go unnoticed. In conclusion, the chosen methodology can explicate mechanisms of action in psychotherapy research and illustrate the enduring patterns underlying PD.