Abstract Objective Various subjective and objective criteria are used to assess outcome in bipolar disorder. In this study, we explored to what extent they reflect distinct categories and whether underlying dimensions can be identified. Patients and methods One-hundred and twenty-one subjects with at least three episodes of bipolar I disorder (DSM-IV) were assessed on average 4.8 years after hospitalization. We assessed 14 variables reflecting different outcome criteria including subjective quality of life (SQOL), self-rated and observer-rated psychopathology, and functioning and disability. A principal component analysis was computed across all outcome variables. Identified dimensions were correlated with sociodemographic characteristics, illness history, premorbid adjustment and personality traits. Results Three outcome dimensions were identified, i.e. a ‘general subjective’, a ’functioning/disability’ and a ‘manic/psychotic symptoms’ dimension. Together they explain 69% of the total variance. The ‘general subjective’ dimension consists of SQOL scales and self-rated depressive symptoms. It is associated with comorbid anxiety disorders and personality disorders, high neuroticism and not having been in hospital in the last year. The ‘functioning/disability’ dimension comprises of criteria reflecting negative symptoms, disability and low functioning. It is associated with more prior illness episodes and low premorbid adjustment. The ‘manic/psychotic symptoms’ dimension consists of observer-rated manic and positive psychotic symptoms. It is correlated with not currently taking a specific medication. Limitations Cross-sectional design with a limited sample size. Conclusion The findings indicate that outcome criteria in bipolar I disorder can be grouped into three distinct dimensions reflecting (1) subjective appraisals, (2) functioning/disability and (3) manic/psychotic symptoms. While measurement of psychotic/manic symptoms has become a matter of course, until now few studies have assessed disability or subjective appraisal in bipolar illness. Therefore important aspects of bipolar illness might be overseen. For a better understanding, we suggest that longitudinal studies of bipolar I disorders should consider all three dimensions of outcome and measure them separately.