Failure to recognize subthreshold expressions of mania contributes to the frequent under-diagnosis of bipolar disorder. There are several reasons for the lower rate of recognition of subthreshold manic symptoms, when compared to the analogous pure depressive ones. These include the lack of subjective suffering, enhanced productivity, ego-syntonicity, and diurnal and seasonal rhythmicity associated with many of the manic and hypomanic symptoms, and the psychiatrists' tendency to subsume persistent or even alternating symptoms among personality disorders. Furthermore, the central diagnostic importance placed on alterations in mood distracts clinicians from paying attention to other more subtle but clinically meaningful symptoms, such as changes in energy, neurovegetative symptoms and distorted cognitions. Although officially accepted in both ICD-10 and DSM-IV, we believe bipolar II disorder is underdiagnosed because of inattention to symptoms of hypomania. Moreover, by requiring the presence of both full-blown hypomanic and major depressive episodes, current nosology fails to include symptoms or signs which are mild and do not meet threshold criteria. There is already agreement in the field that such symptoms are important for depression. We now propose that attention should also be devoted to mild symptomatic manifestations of a manic diathesis, even if such manifestations may sometimes enhance quality of life. The term 'spectrum' is used to refer to the broad range of such manifestations of a disorder from core symptoms to temperamental traits. Spectrum manifestations may be present during, between, or even in the absence of, an episode of full-blown disorder. We have developed a structured clinical interview to assess the mood spectrum (SCI-MOODS) to evaluate the whole range of depressive and manic symptoms. This instrument is currently undergoing psychometric testing procedures. Similar to the SCID interview, the SCI-MOODS interview provides a separate rating for each of the major DSM-IV symptoms, but the latter also identifies and rates subthreshold and atypical manifestations. This paper presents the concept of a subthreshold bipolar disorder and discusses the potential epidemiological, diagnostic and therapeutic relevance of such a spectrum conditions. We also describe the SCI-MOODS interview used reliably to identify the occurrence of a bipolar spectrum condition. Obviously a great deal of systematic research needs to be conducted to ascertain the reliability and validity of subthreshold bipolarity as summarized in this paper and embodied in our instrument.