DSM-IV, specifically its text revision DSM-IV-TR, remains the preferred diagnostic system. When employed in general population samples, prevalence estimates of bipolar disorder are relatively consistent across studies in Europe and USA. In community studies, first onset of bipolar mood disorder is usually in the mid-teenage years and twenties, and the occurrence of a major depressive episode or hypomania is usually its first manifestation. Since reliable criteria for delineating unipolar (UP) and bipolar (BI) depression cross-sectionally are currently lacking, there is a longitudinal risk - probably over 10% - that initial UP patients ultimately turn out as BP in the longer run. Its early onset implies a severe potential burden of disease in terms of impaired social and neuropsychological development, most of which is attributable to depression. BIPOLAR DEPRESSION IN CHILDREN: Bipolar I disorder is rare in prepubertal children, when defined according to unmodified DSM-IV-TR criteria. A broad diagnosis of bipolar disorder risks confounding with other childhood psychopathology and has less predictive value for bipolar disorder in adulthood than the conservative definition. Nevertheless, empirical studies of drug and other treatments and longitudinal studies to assess validity of the broadly defined phenotype in children and adolescents are desirable, rather than extrapolation from adult bipolar practice. The need for an increased capacity to conduct reliable trials in children and adolescents is a challenge to Europe, whose healthcare system should allow greater participation and collaboration than other regions, via clinical networks. ECNP will aspire to facilitate such developments. BIPOLAR DEPRESSION IN ADULTS - UNIPOLAR/BIPOLAR CONTRAST: Despite some differences in symptom profiles and severity measures, a cross-sectional categorical distinction between bipolar (BP) and unipolar (UP) depression is currently impossible. For regulatory purposes, a major depressive episode, meeting DSM-IV-TR criteria, remains the same diagnosis, irrespective of the overall course of the disorder. However, in refining diagnosis in future studies and DSM-V, a probabilistical approach to the UP/BP distinction is more likely to be informative as recommended by the International Society for Bipolar Disorders (ISBD). Anxiety is a commonly present, often at syndromal levels, in bipolar populations. Thus, RCT inclusion criteria for trials not targeting anxiety, should accept co-morbid anxiety disorders as part of the history and even current anxiety symptoms, where these are not dominating the mental state at recruitment to a study. Rapid cycling patients defined as those suffering from 4 or more episodes per year, may also be recruited into trials of bipolar depression without impairing assay sensitivity. Illness severity critically affects assay sensitivity. The minimum scores for entry into a bipolar depression trials should be >20 on HAM-D (17 item scale). However, efficacy is best detected in patients with HAM-D >24 at baseline. THE USE OF RATING SCALES IN BIPOLAR DEPRESSION: There is some dissatisfaction with the HAM-D or MADRS as the preferred primary outcome for trials, although they probably capture global severity adequately. Secondary measures to capture so-called atypical symptoms (such as hypersomnia or hyperphagia), or specific psychopathology more common in bipolar participants (such as lability of mood), could be informative as secondary measures. TREATMENT STUDIES IN BIPOLAR DEPRESSION: Monotherapy trials against placebo remain the gold-standard design for determining efficacy in bipolar depression. The confounding effects of co-medication are emerging from the literature on antidepressant studies in bipolar depression, often conducted in combination with antimanic agents to avoid possible switch to mood elevation. Three arm trials, including the compound to be tested, placebo, and a standard comparator, are generally preferred in order to ensure assay sensitivity and a better picture of benefit-risk ratio. However, in the absence of any gold-standard, two-arm trials may be enough. If efficacy happens to be proven as monotherapy, new compounds may be tested in adjunctive-medication placebo-controlled designs. Younger adults, without an established need for long-term medication, may be particularly suitable for clinical trials requiring placebo controls. The conversion rate of initial UP depression, converting to become BP in the long run is estimated to be 10%. Switch to mania or hypomania may be the consequence of active treatment for bipolar depression. Some medicines such as the tricyclic antidepressants and venlafaxine may be more likely to provoke switch than others, but this increased rate of switch may not be seen until about 10 weeks of treatment. Twelve week trials against placebo are necessary to determine the risk of switch and to establish continuing effects. Careful assessment at 6-8 weeks is required to ensure that patients who are failing to respond do not continue in a study for unacceptable periods of time. To capture a switch event, studies should include scales to define the phenomenology of the event (e.g. hypomania or mania) and its severity. These may be best applied shortly after the clinical decision that switch is occurring. Long-term treatment is commonly required in bipolar disorder. Trials to detect maintenance of effect or continued response in bipolar depression should follow a 'relapse prevention' design: i.e. patients are treated in an index episode with the medicine of interest and then randomized to either continue the active treatment or placebo. However, acute withdrawal of active medication after treatment response might artificially enhance effect size due to active drug withdrawal effects. A short taper is usually desirable. Longer periods of stabilisation are also desirable for up to 3 months: protocol compliance may then be difficult to achieve in practice and so will certainly make studies more difficult and expensive to conduct. The addition of a medicine to other agents during or after the resolution of a depressive or manic episode, and its subsequent investigation as monotherapy against placebo to prevent further relapse (as in the lamotrigine maintenance trials) is clinically informative. Assay sensitivity and patient acceptability are enhanced if the outcome in long-term studies is 'time to intervention for a new episode' for discontinuation designs.