Continuation of antidepressant therapy after patients have responded to medication has been shown to greatly reduce but not eliminate relapse. We evaluated response to reinstating or increasing drug doses in patients who relapsed during a long-term, randomized, double-blind, continuation phase study of duloxetine treatment for major depressive disorder (MDD). Patients with HAMD17 scores of >18 and CGI-Severity scores of >4 received duloxetine 60 mg QD for 12 weeks, responders then received 26 weeks of duloxetine 60 mg QD or placebo. The relapsing patients who started 12 weeks of rescue treatment received duloxetine 60 mg QD (if relapsing on placebo) or duloxetine 60 mg BID (if relapsing on duloxetine). In the continuation phase, 87 patients received duloxetine 60 mg QD (n=58) or 60 mg BID (n=29) as treatment for relapse. The percentage of patients who responded to treatment after relapse was 62% of those whose dose was increased from 60 mg QD to 60 mg BID, and 74% for those who switched from placebo to duloxetine 60 mg daily. By the end of the study, 32 patients taking duloxetine 60 mg QD were in remission (57%), compared with 11 patients taking duloxetine 60 mg BID (38%). There was significant improvement in HAMD17 and CGI-Severity scores for both groups at endpoint vs. start of treatment for relapse. Significant improvements were seen for all visual analog scale parameters in the 60 mg QD group, but only Interference with Daily Activities and Back Pain parameters showed significant improvement for patients receiving 60 mg BID. Significant within-group improvement from baseline was seen for both groups in Symptom Questionnaire-Somatic Subscale total and pain subscales. Treatment-emergent adverse events were mild/moderate in severity; no clinically significant changes were noted in laboratory results or vital signs. Reinstatement of duloxetine 60 mg QD was effective for patients who relapsed after discontinuing drug. Patients relapsing on duloxetine 60 mg QD benefited from an increase to 60 mg BID. These duloxetine doses were well tolerated and effective, and appear appropriate for MDD patients requiring treatment of relapse.