Background Individuals with bipolar disorder lead a sedentary lifestyle associated with worse course of illness and recurrence of symptoms. Identifying potentially modifiable predictors of exercise frequency could lead to interventions with powerful consequences on the course of illness and overall health. Methods The present study examines baseline reports of exercise frequency of bipolar patients in a multi-site comparative effectiveness study of a second generation antipsychotic (quetiapine) versus a classic mood stabilizer (lithium). Demographics, quality of life, functioning, and mood symptoms were assessed. Results Approximately 40% of participants reported not exercising regularly (at least once per week). Less frequent weekly exercise was associated with higher BMI, more time depressed, more depressive symptoms, and lower quality of life and functioning. In contrast, more frequent exercise was associated with experiencing more mania in the past year and more current manic symptoms. Limitations Exercise frequency was measured by self-report and details of the exercise were not collected. Analyses rely on baseline data, allowing only for association analyses. Directionality and predictive validity cannot be determined. Data were collected in the context of a clinical trial and thus, it is possible that the generalizability of the findings could be limited. Conclusion There appears to be a mood-specific relationship between exercise frequency and polarity such that depression is associated with less exercise and mania with more exercise in individuals with bipolar disorder. This suggests that increasing or decreasing exercise could be a targeted intervention for patients with depressive or mood elevation symptoms, respectively. © 2013 Elsevier B.V.