Register-based cohorts allow us to better understand bipolar disorder over a life course. They are inclusive and their long-term data collection provides a longer scope than most clinical trials. This mapping review provides an overview of register-based cohort studies of bipolar disorder to inform researchers of the strengths and limitations to this body of research and identify gaps for future research. A systematic search was performed of Medline, EMBASE, and PsycINFO databases. Cohort studies were included if they focused on bipolar disorder and had a minimum of 1 year of longitudinal data. Studies needed to be from databases that monitor the whole state or national population. A descriptive analysis of the studies' populations and methodology provides an overview of this field of study and identifies evidence gaps. A hundred and forty-six studies were included. The majority were from databases in Taiwan (n = 63), Denmark (n = 38), Sweden (n = 23), and Finland (n = 11). Forty-eight studies focused on aetiological questions. Sixty prognostic studies identified cohorts with bipolar disorder and described the impact of the illness by considering comorbidity, prescribing patterns, social functioning, and mortality. Thirty-six treatment studies focused on the efficacy and adverse effects of pharmaceuticals and ECT. No studies focused on psychological treatments. Bipolar disorder research should include register-based cohorts with greater geopolitical and cultural diversity. Custodians of health registers should consider how non-pharmaceutical interventions such as psychotherapy are captured. Register-based cohorts investigating treatments of bipolar disorder should consider long-term social outcomes alongside the usual clinical outcomes.