• Gunderson's Diagnostic Interview for Borderlines (DIB) has gained increasing acceptance and use as a diagnostic research instrument. Although previous studies report interrater reliability for the DIB, no attention has been given to problems of test-retest reliability or variability among different clinicians conducting a semistructured clinical interview. This study demonstrated both interrater and test-retest reliability for the DIB when it is administered by trained clinicians. No effects of interviewer sex were found, but less experienced clinicians tended to give higher (more borderline) scores than more experienced clinicians.