The authors report preliminary findings from a longitudinal study on the impact of attachment state of mind and reflective function on therapeutic process and outcome with borderline patients in Transference-Focused Psychotherapy (TFP). TFP is a manualized, psychoanalytically oriented treatment based on an object relations model of understanding patients with severe personality disorders. The attachment theory constructs of internal working models of attachment and mentalization or reflective function provide an important means of both conceptualizing borderline disorders and assessing therapeutic process and change. In the Personality Disorders Institute at New York Presbyterian Hospital-Weill Medical College of Cornell University, the authors have been using the Adult Attachment Interview (AAI) to assess changes in state of mind with respect to attachment and reflective function over the course of 1 year in borderline patients in TFP treatment. As part of the authors' investigations of the impact of patients' attachment status on the therapeutic process, they have adapted the AAI to evaluate states of mind with respect to attachment within the therapeutic relationship through an interview called the Patient-Therapist Adult Attachment Interview (PT-AAI). The AAI is given at 4 months and 1 year, and the PT-AAI is given to patients after 1 year of TFP, and both interviews are scored for attachment classification and reflective function. The authors present preliminary findings on change in both attachment classification and reflective function ratings at 4 months and 1 year for a subsample of 10 patients and therapists. They also present two cases that illustrate how the quality of mentalization or reflective function in the therapeutic dyad may be seen as a bidirectional process in that therapists' and patients' levels of reflective function are mutually and reciprocally influential. In one case, the patient's and therapist's reflective function mirrored each other directly and remained at a low or rudimentary level for the treatment year. Such a pattern of direct imitation does not necessarily promote intrapsychic change. In the second case, the patient moved from a rejecting or bizarre stance toward mentalization on the AAI to some rudimentary consideration of mental states after 1 year of treatment with a therapist who showed a full and nuanced awareness of mental states, but who adjusted his level of mentalization to that of the patient. These findings suggest that optimally the therapist ought to be one step ahead of the patient in the capacity for mentalization.