Bipolar (manic-depressive) disorder is one of the most common of the severe mental illnesses. Officially recognised forms comprise type I (with mania), type II (with hypomania), cyclothymia and a rapid-cycling subtype. International lifetime prevalence estimates are 1 to 5% of the general population, and bipolar disorder accounts disproportionately for idiopathic psychoses. Psychiatric and substance-abuse comorbidities are common complications, and mortality rates are increased as a result of high suicidal risks, accidents, complications of substance abuse and increased fatality of stress-sensitive medical illnesses. Complex and labile symptomatic presentations, a tendency for patients to deny illness and reject treatment, and diagnostic heterogeneity severely complicate the design, conduct and interpretation of experimental treatment trials in bipolar disorder.Progress in the short-term treatment of mania with certain antiepileptic drugs and atypical antipsychotic agents has advanced greatly in recent years; however, long-term treatment trials other than with lithium remain rare, as are studies of type II disorder, bipolar depression and mixed states, and there is limited information on treatment effectiveness against comorbidity, dysfunction and mortality. There is a growing realisation that bipolar disorder represents a major, largely unmet, international public health challenge and that innovative methods for carrying out reliable and generalisable long-term pharmacological treatment trials, alone and in combination with cost-effective psychosocial and rehabilitative interventions, are urgently required.