Abstract: Background: The multiple factors likely to influence estimates of the prevalence of bipolar spectrum disorders (BSD) make it difficult to determine its actual prevalence. A systematic review was undertaken to explore the availability and quality of epidemiological data for BSD. This will inform the Global Burden of Disease 2010 study.Methods: Electronic databases searched were Medline, PsycInfo and EMBASE using search terms generated in consultation with a research librarian. Reference lists were searched and experts contacted to obtain articles not identified through the database search. Prevalence data pertaining to point, 6month or 12month prevalence of BSD were sought.Results: The mean pooled prevalence was 0.741% (point) and 0.843% (6/12months). The study region, response rate and diagnostic tool had a significant impact on prevalence estimates but there were no significant differences in prevalence across prevalence types, gender, sample coverage, economic status and bipolar subtype. Little or no prevalence data was apparent for many regions of the world. Also, the entire bipolar spectrum was rarely assessed.Limitations: Although stringent quality assurance methodology was used, controlling for all sources of variability around the pooled prevalence was not possible.Conclusions: This systematic review has made significant contributions to the epidemiological profile of BSD. Prevalence estimates will be used to generate estimates of burden for BSD in the global burden of disease 2010 study.