Bipolar disorder is a multifactorial psychiatric disorder with developmental and progressive neurophysiological alterations. This disorder is typically characterized by cyclical and recurrent episodes of mania and depression but is heterogeneous in its clinical presentation and outcome. Although the DSM-IV-TR criteria identify several features that are of phenomenological relevance, these are of less utility for defining homogeneous subgroups, for analyses of correlations with biomarkers or for directing focused medication strategies. We provide a comprehensive review of existing evidence regarding to age at onset in bipolar disorder. Eight admixture studies demonstrate three homogeneous subgroups of patients with bipolar disorder identified according to age at onset (early, intermediate and late age at onset), with two cutoff points, at 21 and 34 years. It is suggested that the early-onset subgroup has specific clinical features and outcomes different from those of the other subgroups. Early-onset subgroup may be considered a more suitable clinical phenotype for the identification of susceptibility genes with recent data demonstrating associations with genetic variants specifically in this subgroup. The use of age at onset as a specifier may also facilitate the identification of other biological markers for use in brain imaging, circadian, inflammatory and cognitive research. A key challenge is posed by the use of age at onset in treatment decision algorithms, although further research is required to increase the evidence-base. We discuss three potential benefits of specifying age at onset, namely: focused medication strategies, the targeted prevention of specific comorbid conditions and decreasing the duration of untreated illness. We argue that age at onset should be included as a specifier for bipolar disorders.