Clinicians need to know whether duloxetine is effective in patients across a broad range of depressive symptoms and depression severity. Data were pooled from nine randomized, double-blind, placebo-controlled studies in major depressive disorder (total N=2227) comparing duloxetine (40-120 mg/day) with placebo for 8-9 weeks. Patients were retrospectively stratified by baseline score on the HAMD17 into mild (≤19; n=682), moderate (n=1099), or severe (≥25; n=446) groups. Duloxetine produced significantly greater baseline-to-endpoint mean change than placebo in HAMD17 total score, Maier and retardation subscales, and the Clinical Global Impressions-Severity of Illness scale in all three cohorts. Significant improvement was seen in HAMD17 items 1 (depressed mood), 3 (suicide), 7 (work and activities), and 10 (psychic anxiety) regardless of severity. The HAMD17 anxiety subscale and items 13 (somatic symptoms-general) and 15 (hypochondriasis) showed significant improvement only in moderately and severely ill patients. Significant improvement in the HAMD17 Maier subscale was seen in all groups by week 1. In all three groups, placebo was significantly superior to duloxetine at early visits on HAMD17 item 12 (somatic symptoms-GI). Mildly and severely ill patients exhibited significant reduction in visual analog scale overall pain severity at the study endpoint. The studies contained fewer patients with very mild or very severe illness, limiting our ability to draw conclusions in these patient populations. Duloxetine demonstrated superior efficacy in the treatment of major depressive disorder, when compared with placebo, regardless of the baseline severity of depressive symptoms, although effect sizes were largest in the most severely depressed patients. © 2007 Lippincott Williams & Wilkins, Inc.