OBJECTIVES: Existing research offers little empirical evidence to support national practice guidelines about antidepressant use in patients with bipolar disorder. This study assessed the clinical and economic impact of second-generation antidepressant use for bipolar depressed adult patients. METHODS: Bipolar disorder subjects were identified with a new depressive episode in a national managed-care plan between January, 1998 and December, 2002. In total, 3737 patients were defined with the index of interested medication initiation and at least three-months pre-index and 12-month post-index continuous enrollments. Logit model with a difference-in-difference approach was employed to identify the relationship between treatment types (antidepressant monotherapy, mood stabilizer monotherapy, and antidepressant-mood stabilizer combination therapy) and the probability of mania-related visits. In addition, negative binomial model and log-transformed OLS model were used to predict number of depression-related visits and healthcare costs respectively. RESULTS: Probability of mania-related visits increased significantly from pre-index to post-index period with odds ratio 2.40 (95% CI: 1.52-3.79) for antidepressant monotherapy compared to mood stabilizer monotherapy, controlling for time fixed-effect, demographics, clinical-related and health-related variables. However no significant difference was identified between combination therapy and mood stabilizer monotherapy. Patients with antidepressant monotherapy and combination therapy had significantly lower incidence rate ratios of 0.68 (95% CI: 0.56-0.82) and 0.65 (95% CI: 0.52-0.81) respectively for depression-related visits, but significantly higher bipolar-related costs (232% and 72% respectively) compared to patients with mood stabilizer monotherapy. CONCLUSIONS: This study adds to the literature of bipolar depression treatment by providing empirical evidence at the national level to support the current practice guidelines. Second-generation antidepressant monotherapy indicated a risk of induced manic-switching, which can be costly in terms of overall healthcare resources. Since the safety and efficacy of antidepressants in bipolar disorder have not been firmly established, medical practitioners and policy makers should take precautions regarding the appropriate use of new antidepressants.