High rate of relapse and inter-episode dysfunction necessitate various attempts to enhance the outcome of the maintenance therapy of bipolar disorder. The aim of this naturalistic follow-up study was to explore the effects of aripiprazole adjunctive therapy on relapse prevention in patients with bipolar disorders. Patients with DSM-IV bipolar I or II disorder were treated with adjunctive aripiprazole combined with mood stabilizers and followed up. The primary endpoint was time to relapse for mood episode. To investigate the relationship between the polarity of recent episode and time to relapse, the statistical difference in Kaplan-Meier estimates between recently manic and depressive groups was tested using log-rank method. The number of episode per year before and after the initiation of aripiprazole add-on were also calculated. A sigficantly longer delay in time to manic relapse was observed in recently manic group compared to recently depressive group. No significant association was observed between delay in time to depressive relapse and polarity of recent episode. Adjunctive usage of aripiprazole led to a significant reduction in the number of episode per year. These data lend support to the effectiveness of long-term adjunctive aripiprazole treatment in the prevention of manic relapse with weaker protection against depressive relapse. Polarity of recent episode appeared to be associated with manic relapse. Further controlled trials are needed to confirm these findings.