Background: Young adults with early-onset major depressive disorder (MDD) may be at high risk of progression to bipolar disorder. Although hypomanic symptoms are common in young people with depression, many do not reach the strict DSM-IV and ICD-10 criteria for hypomania. We used an emerging innovative framework for bipolar spectrum to evaluate this question. Methods: Consecutive referrals to a psychiatric outpatient clinic at a university health service were assessed for recurrent episodes of depression. DSM-IV diagnoses were based on a SCID-1 interview. We used two approaches to delineate bipolar spectrum. The first focused on bipolar spectrum disorder (BSD, as defined by Ghaemi et al. [Can. J. Psychiatry 47 (2002) 125]), and the second on a symptoms perspective based on MDD with a history of hypomanic symptoms, using a 15-point hypomanic symptoms checklist with a cut-off ≥8 or more symptoms (modified from J. Affect. Disord. 73 (2003) 39 and J. Affect. Disord. 73 (2003) 73). Data were also obtained on family history of affective disorder, course and number of episodes of depression, symptom severity, psychosocial functioning, suicidality and deliberate self-harm, and drug and alcohol use. Results: High rates of bipolar and bipolar spectrum disorder were identified. Under DSM-IV, 14 subjects (16.1%) had bipolar affective disorder and 73 subjects (83.9%) had recurrent MDD. Depending on the method used to diagnose bipolar spectrum, between 47.1% and 77.0% of the total cohort could be so diagnosed. Hypomanic symptom counts, irrespective of duration, yielded the highest estimates for bipolar spectrum. High rates of pharmacological hypomania were also identified: 12 subjects (16.4%) with recurrent MDD group reported this, and all could be diagnosed with bipolar spectrum. Limitations: The reliability of using the 15-point hypomanic scale for the diagnostic assignments was not tested. All subjects were recruited from a university health service and, given the affluence of their parents, findings may not generalise to other populations. Most importantly, because bipolar family history and pharmacological hypomania were part of the diagnostic criteria of the BSD group, they could not be used as external validators for Ghaemi's BSD construct. Conclusions: Bipolar disorders emerge as extremely common in this cohort of young adults with recurrent depression. Antidepressant-induced hypomania and high scores on a hypomanic symptoms checklist help to identify patients who are likely to have a bipolar spectrum illness, but who do not meet DSM-IV criteria for bipolar disorder. This is a preliminary study, and further evidence from external validating strategies are needed to verify the bipolar status of these patients in a larger and unselected cohort representing a broader socio-economic demographic profile.