The diagnosis and treatment of bipolar disorder (BD) has been inconsistent and frequently misunderstood in recent years. To identify the causes of this problem and suggest possible solutions, we undertook a critical review of studies concerning the nosology of BD and the effects of antidepressant agents. Both the underdiagnosis of BD and its frequent misdiagnosis as unipolar major depressive disorder (MDD) appear to be problems in patients with BD. Underdiagnosis results from clinicians' inadequate understanding of manic symptoms, from patients' impaired insight into mania, and especially from failure to involve family members or third parties in the diagnostic process. Some, but by no means all, of the underdiagnosis problem may also result from lack of agreement about the breadth of the bipolar spectrum, beyond classic type I manic-depressive illness (what Ketter has termed "Cade's Disease"). To alleviate confusion about the less classic varieties of bipolar illness, we propose a heuristic definition, "bipolar spectrum disorder." This diagnosis would give greater weight to family history and antidepressant-induced manic symptoms and would apply to non-type I or II bipolar illness, in which depressive symptom, course, and treatment response characteristics are more typical of bipolar than unipolar illness. The role of antidepressants is also controversial. Our review of the evidence leads us to conclude that there should be less emphasis on using antidepressants to treat persons with this illness.