This study will compare the effectiveness of relatively new antidepressants which have different mechanisms of action.

Buproprion (Wellbutrin) works on dopamine and the dopaminergic pathway.

Sertraline (Zoloft) works as a selective serotonin reuptake inhibitor (SSRI).

Venlafaxine (Effexor) works as a mixed serotonin, norepinephrine, and dopamine reuptake inhibitor.

Subjects enrolled in this study will be patients diagnosed with a bipolar disorder who are presently taking medication to prevent the symptoms of the disease (prophylactic treatment), but have had breakthrough episodes of depression despite taking their medication.

Patients will receive any one of the three antidepressant medications as noted above plus a placebo inactive sugar pill, in order to mask which antidepressant is being prescribed) in addition to their regular medication for bipolar disorder. All of the doses will be calculated as effective for the treatment of a unipolar major depressive disorder. The patient will continue receiving the medication for ten weeks.

The effectiveness of the drug treatment will be measured by using three different scales;

1. Inventory for Depressive Symptoms - Clinicians form (IDS-C)
2. Clinical Global Impression scale(CGI-BP)
3. Life Charting Methodology (LCM)

Patients who do not respond to their medication within ten weeks from the beginning of the study will be considered as non-responders and be offered the opportunity to start the study again, taking one of the two remaining medications. For example, if a patient was assigned to take Wellbutrin but it was ineffective, he/she could re-enter the study and be given either Zoloft or Effexor.

Patients that do respond in the first ten weeks of the study will be eligible to continue taking the medication for one year to assess the long term effectiveness of the drug on preventing episodes of depression and to assess for any possible differential induction of mania.

This NIMH-Stanley Foundation Bipolar Network (the "Network") study will be the first systematic assessment in bipolar depression of the comparative efficacy of bupropion (Wellbutrin), sertraline (Zoloft), and venlafaxine (Effexor), three newer antidepressants which have very different mechanisms of action. Bupropion is largely dopaminergic while sertraline is a serotonin selective reuptake inhibitor (SSRI) and venlafaxine is a mixed serotonin, norepinephrine, and to a lesser extent dopamine reuptake inhibitor. Subjects enrolled in this study will be bipolar patients on prophylactic treatment who experience a breakthrough major depressive episode. Subjects will be assigned in a double-blind manner using a three-arm forced randomization procedure to antidepressant therapy with either bupropion, sertraline, or venlafaxine for a ten-week acute response trial. The study medications will be assigned as adjuvant treatment to mood stabilizing medication(s) which have proven unsatisfactorily effective within therapeutic range(s) or at maximum tolerated dose(s). All subjects will receive active drug at dosages established as clinically relevant for unipolar major depressive disorder. Subjects and research personnel conducting cross-sectional and longitudinal rating assessments of mood and functioning will be blinded to treatment group assignment. The primary outcome measures will be the Inventory for Depressive Symptoms - Clinician form (IDS-C), the Clinical Global Impression (CGI-BP) scale, and the Life Charting Methodology (LCM). Subjects who worsen from baseline to week four of treatment will be considered non-responders to the initially assigned medication, and will be offered re-randomization to either of the two drugs to which the subject was not originally randomized (e.g., for bupropion non-response, to either sertraline or venlafaxine) for an additional ten-week acute response trial. Responders to the acute trial enter a 12 month trial of continuation therapy to assess long term effects on prophylaxis of depression and possible induction of mania or cycle acceleration. We also wish to explore possible clinical and biological correlates of acute and long term response to these antidepressants which have received almost no systematic study in bipolar illness. One hypothesis of this study is that the acute efficacy for the three antidepressants would be the same. Another hypothesis is that because of the potent noradrenergic (NE) effects, venlafaxine would have a higher rate of inducing mania compared with bupropion or sertraline.

Subjects fulfill DSM-IV criteria for Bipolar I disorder (BPI), Bipolar II disorder (BPII), Bipolar disorder not otherwise specified (BPNOS), or schizoaffective disorder bipolar type.

Subjects must be competent to comprehend the purpose of the study and provide informed consent.

Subjects must undergo complete psychiatric diagnostic interview (SCID--DSM-IV), medical, neurological, and Laboratory examinations (including EKG, renal and liver function tests, serum electrolytes, urinalysis, HIV, hepatitis B, pregnancy testing, and urine drug screen for the presence of psychoactive drugs and drugs of abuse).

At least 18 years old.

Subjects must have a depression of sufficient severity to rate greater than or equal to 16 on the Inventory of Depressive Symptomatology -Clinician (IDS-C) comparable to greater than or equal to 12 on the Hamilton Depression Rating Scale) or the clinician must decide that there is a need to treat with an antidepressant. In addition, patients must be on at least one mood stabilizer.

Subjects should have no general medical illness that is causing the mood disorder.

Subjects should not have liver, renal, hematological, or neurological disease.

Women participants of childbearing potential must be nongravid, nonnursing, and using an acceptable method of birth control.

Patients must not have alcohol or substance use or dependence of sufficient magnitude to require independent, concurrent treatment intervention (excluding self-help groups), i.e., hospitalization, day treatment programs, or counselor visits.

No patients taking concomitant medications that would contraindicate the medications under study, such as chemotherapy.

No history of bulimia or seizure disorder.