OBJECTIVE: Investigate the effect of atypical antipsychotics (quetiapine, risperidone, and olanzapine) in combination with a mood stabilizer on hospitalization rates for bipolar disorder. METHOD: From the MEDSTAT MarketScan(c) medical claims database (1998 - 2001), 977 individuals were identified who had a diagnosis of bipolar disorder and received combination therapy with an atypical antipsychotic and a mood stabilizer. A 2-stage sample selection model controlled for differences between individuals receiving antipsychotics and factors that may impact the probability of hospitalization, i.e. demographics, bipolar type, disease severity, comorbidities, mood stabilizer used, and antipsychotic used. RESULTS: Individuals most likely to be hospitalized were those diagnosed as manic or bipolar depressed, or were receiving divalproex sodium or gabapentin. In pair-wise comparisons, hospitalization in the year following the start of combination antipsychotic therapy was 44% less likely for those receiving quetiapine vs olanzapine (P = 0.0354). There was no significant difference in the likelihood of hospitalization for the quetiapine group compared to the risperidone group (P = 0.5826). CONCLUSIONS: In patients with bipolar disorder receiving a mood stabilizer plus an atypical antipsychotic, the probability of hospitalization with quetiapine was significantly lower than with olanzapine, and similar to risperidone. These findings are relevant to prescription choices among atypical antipsychotics, for maximizing patient benefit and minimizing the burden of disease.