Background: Anecdotal reports concerning mood disorder patients suggest restarting drug treatment in patients who prematurely discontinue it may be associated with a diminished chance of response upon rechallenge. We evaluated the likelihood of response to reinitiation of fluoxetine treatment in patients relapsing after switching to placebo during a long-term efficacy study of two different dosing regimens of fluoxetine. Method: Patients who met the DSM-IV major depressive disorder criteria with modified HAMD17 scores ≧18 and CGI-Severity scores ≧4 were treated with open-label 20 mg/day fluoxetine for 13 weeks in a multicenter US study. Responders (n = 501) were randomized to placebo, 20 mg fluoxetine daily, or 90 mg enteric-coated fluoxetine weekly for 25 weeks of double-blind continuation treatment. Patients relapsing during the continuation phase were offered 25 weeks of double-blind rescue treatment during which the study medication dose was increased as follows: (1) patients on placebo had treatment with fluoxetine 20 mg/ day reinitiated; (2) patients on fluoxetine 20 mg/day had their dose increased to 40 mg/day, and (3) patients on a 90-mg weekly dose had their dose increased to 90 mg twice a week. Only the results of the rescue phase for the group who relapsed while on continuation treatment with placebo are reported here. Analyses included percentage of responders (50% reduction of modified HAMD17 and CGI-Severity ≤2) and assessments of change from baseline to endpoint (HAMD, CGI-Severity). Safety measures included assessment of vital signs, laboratory measures, and treatment-emergent adverse events. Results: Of 122 patients assigned to placebo, 57 (47%) relapsed during continuation treatment. Fifty-five (96%) of these elected to enter double-blind rescue treatment. Overall, patients who relapsed upon switching to placebo responded well to reinitiation of fluoxetine (62%). The mean modified HAMD17 score decreased from 20 to less than 9 and was maintained for up to 6 months. Thirty-eight percent of patients either did not respond or initially responded but again relapsed after reinitiation of medication. Conclusion: This study suggests that patients who, after an initial response to fluoxetine, relapse upon switching to placebo have a relatively high probability of responding to the reinitiation of fluoxetine treatment. These results challenge the view that the efficacy of an agent prematurely discontinued is diminished when such an agent is restarted. They also generally support reinitiation of the same antidepressant as a ‘first-line’ treatment strategy in patients who relapsed after stopping a previously effective antidepressant.