After three decades of clinical controversy and research, a clinical consensus has formed that borderline personality disorder (BPD) exists as a unique entity which can be defined by DSM-III-R diagnostic criteria. The purpose of this study was to evaluate the relative abilities of four different approaches to the identification of borderlines and to differentiate a DSM-III-R BPD group from a control group of other diagnoses. The approaches were the Kernberg's Structural Interview, the Diagnostic Interview for Borderline Personality Disorders (DIB), the Borderline Syndrome Index (BSI), and the Million Clinical Multiaxial Inventory (MCMI). Sixty outpatient volunteers (27 men and 33 women) from a community mental health center served as subjects. The volunteers included 30 BPDs and 30 other diagnoses, including 11 non-BPD personality disorders. Point biserial correlations indicated that the best method for identifying DSM-III-R BPD was the DIB, and the second best was the Kernberg Structural Interview, although all four identified DSM-III-R BPDs at better than chance levels. Multiple regression results showed that the DIB accounted for 61.5% of the BPD variance, while the Kernberg approach added 4.9% more unique variance prediction. MCMI dimensions and personal history characteristics were used to identify differentiators of BPD from all other diagnoses and from other personality disorders.