BACKGROUND In young adults it can be difficult to differentiate between an early bipolar illness and borderline personality disorder. There are considerable areas of clinical overlap between cyclothymic temperament, bipolar-spectrum disorders and borderline characteristics. The aim of this study was to measure borderline characteristics in young adults during an index depressive episode and to compare three diagnostic groups: DSM-IV bipolar affective disorder (BPAD); bipolar spectrum disorder (BSD); and DSM-IV recurrent major depressive disorder (MDD). METHODS Eighty-seven young adults with a current episode of major depression and at least one previous episode of depression were recruited from consecutive referrals to a psychiatric clinic. Diagnoses were based on the Structured Clinical Interview for DSM-IV (SCID-1) and recently proposed structured diagnostic criteria for BSD. All patients also completed the borderline questions from the screening questionnaire of the International Personality Disorders Examination (IPDE). RESULTS Diagnostically, the cohort of 87 patients divided into three groups: 14 with BPAD; 27 with BSD; and 46 with MDD. None of the subjects fulfilled DSM-IV or ICD-10 diagnostic criteria for personality disorder and all three groups were well matched in terms of age, gender distribution, ethnicity, socioeconomic and educational status, age at onset of illness, and severity of index depressive episode. Both of the bipolar-depressed groups reported significantly higher median levels of borderline characteristics than the MDD group (p<0.0001). Three of the borderline characteristics emerged as potentially useful in differentiating bipolar depression from unipolar depression: 'I've never threatened suicide or injured myself on purpose' (sensitivity=0.93; positive predictive value [PPV]=56.7); 'I have tantrums or angry outbursts' (sensitivity 0.66; PPV=65.6%); and 'Giving in to some of my urges gets me into trouble' (sensitivity=0.76; PPV=59.6%). CONCLUSIONS All of the subjects were recruited from a university health service clinic and as such are unlikely to be representative of patients from more diverse socio-economic backgrounds. No structured diagnostic assessment of personality disorder was administered. The diagnostic criteria for BSD are not yet fully validated. CONCLUSIONS Young adults with bipolar depression exhibit significantly higher levels of borderline personality pathology than those with unipolar depression. Those borderline screening questions that reflect cyclothymic characteristics or depressive mixed states may be of practical use to clinicians in helping to differentiate between bipolar depression and unipolar depression in young adults.