INTRODUCTION: Bipolar disorder (BD) is a life course illness; and there is increasing awareness of the many personal, social and economic consequences of the illness in older adults. However, it is important to emphasize that BD usually begins in late adolescence or early adulthood and 75 % cases have a first episode in this age period. This early onset and the associated level of disability mean that BD is the 4th leading cause of global disease burden in adolescents and young adults. Internationally, mental health services are increasingly striving to diagnose and treat BD as early as possible to try to prevent poor outcomes. In addition, researchers are using methods employed previously in psychosis studies as these may help us to recognise the earliest manifestations of BD. If it is possible to identify sub-threshold and 'ultra high risk' syndromes for BD, this might lead to new interventions that could target the prevention of first episodes of mania. One approach to understanding these risk syndromes is to examine prospective community cohort studies and BD offspring studies. METHODS: This paper reviews prospective cohort studies that identify robust risk factors in early illness onset, which was defined as age at onset of BD between 15-25 years. RESULTS: We found that although > 50 % of individuals who developed BD had developed a putative BD prodrome prior to 14 years of age, this usually began with non-specific symptoms that overlap with similar presentations for those who later develop psychosis or severe depression. However, there are some features that seem to better identify groups with a BD "at-risk" syndrome. This syndrome is frequently composed of several factors such as mood lability, depressive episodes, prior anxiety, sleep and/or conduct disorders, attention and concentration impairment, altered energy patterns, and a family history of mania and/or depression. The course of these early predictors suggests the precursor syndromes are composed of mini-clusters of symptoms many of which are episodic and change over time. During the early phases of BD, most of the affective disturbances reported were depressive in polarity and started during adolescence, there were few manic or mixed or psychotic episodes with an onset before puberty. The pathogenesis of BD demonstrates a gradual progression from non-specific to more specific symptoms and then to frank BD features. CONCLUSION: Prospective community and offspring BD cohort studies are approaches that together can help us understand the evolution of BD and allow us to define the developmental pathways. Further, identifying subjects with BD "at-risk" syndrome using a clinical staging model may allow benign interventions to be used as first-line treatment - such as neuroprotective agents like essential fatty acids; second line treatments, with a less benign risk to benefit ratio should be reserved for severe or resistant cases.