1. As many as 30% of depressed patients fail to respond to antidepressant drug therapy, and at least 60%-75% will not achieve complete recovery. Recently, several studies have suggested that newer, "second generation" antidepressants might be beneficial in treatment resistant depression (TRD). In the present study the authors examined the potential utility of fluoxetine in TRD by comparing its efficacy in patients with and without a prior history of antidepressant response. 2. 149 patients with MDD received fluoxetine 20mg daily for a minimum of 5 weeks: 43 (29%) had never responded to drug therapy (TRD patients), 41 (28%) did have a prior drug response (non-TRD patients) and 65 (44%) had never received any antidepressant treatment. Clinical response was defined as a > or = 50% reduction in baseline Hamilton Depression Rating score plus a final score < 7. 3. Compared to non-TRD patients, the TRD patients were more likely to have unipolar depression (p = 0.002), a chronic episode of > 2 years duration (p < 0.0001), a later age of illness onset (p < 0.0001), fewer prior episodes (p < 0.0001) and fewer prior drug treatments (p = 0.04). Overall, the response rate to fluoxetine was slightly greater in the non-TRD patients (76%) compared to the TRD patients (56%); however, this difference did not achieve statistical significance. 4. The present observations suggest that fluoxetine treatment of adequate duration may be beneficial for some patients with a prior history of refractory depression.