Some evidence suggests that medications that modulate both serotonin and norepinephrine may be more effective than selective serotonin-reuptake inhibitors (SSRIs) in severe major depressive disorder (MDD). This prospective pragmatic trial tests this hypothesis. Patients with severe MDD were randomly assigned to either duloxetine (a serotonin and norepinephrine-reuptake inhibitor) or physicians’ choice of four generic SSRIs. Nonblinded, flexibly dosed treatment was used to mimic clinical practice. To address potential investigator bias, the patient-reported Quick Inventory of Depressive Symptomatology Self-Report (QIDS-SR) was used as the primary efficacy outcome measure. A total of 750 outpatients (19.2%, African descent; 14.8%, Hispanic) were randomized. The primary outcome, remission at week 12 by QIDS-SR, was numerically greater for duloxetine compared with SSRIs (36 vs. 32%), but this difference was not statistically significant. Mean changes in secondary outcomes were significantly superior in favor of duloxetine for the Hamilton Depression Scale-17 item, the Brief Pain Inventory, and the Sheehan Disability Scale. Remission superiority on the QIDS-SR was not achieved. Significantly greater benefit for duloxetine compared with SSRIs was demonstrated on measures of pain and functioning. Study demographics suggest a more generalizable racial and ethnic population than is typical in randomized clinical trials.