The psychometric properties and validity of the Borderline Personality Disorder Severity Index (BPDSI), a semistructured interview assessing the frequency and severity of manifestations of Borderline Personality Disorder (BPD) during a circumscribed period, were investigated in two studies. In study 1, patients with BPD (n = 15), with other personality disorders (PD; n = 18), and without Axis II disorders (but with Axis I disorders; n = 10) assessed with the SCID were interviewed with the BPDSI (1-yr. version). Patients also filled out a number of questionnaires. A second rater judged taped BPDSI interviews. The BPDSI appeared to yield highly reliable (ICC = .93) and internally consistent (Cronbach's alpha = .85) scores. The BPDSI strongly discriminated BPD patients from other patients, and was not related to other Axis II pathology. Concurrent and construct validity was excellent. In study 2, a version of the BPDSI suitable for use in treatment outcome research was investigated (3-month version) in a sample of 64 BPD patients, 23 Cluster C PD patients, and 20 nonpsychiatric controls. Again, reliability coefficients were excellent (ICC = 0.97; Cronbach's alpha = 0.93), and validity indices were good. Clinical norms were also derived. In a sample of 28 BPD patients, the instrument detected improvement during 6 months of psychotherapy.