There are few controlled studies evaluating the treatment of bipolar mixed states. Evidence suggests that mixed states may be more responsive to some anticonvulsants than to lithium. Olanzapine alone or in combination with divalproate or lithium has been adequately evaluated in randomized clinical trials involving mixed-state patients, whereas risperidone and quetiapine have not. There is also some evidence demonstrating the efficacy of ziprasidone and aripiprazole. The risk of switching to depression is high in mixed states. Conventional antipsychotics, such as haloperidol, may be less efficacious at protecting against a switch to depression than atypical antipsychotics, divalproate or lithium. When choosing drugs for the treatment of mania, and especially for the treatment of mixed states, their efficacy against manic and depressive symptoms, and their safety in terms of the risk of switching to depression should be taken into account.