Major depressive disorder (MDD) is characterized by a constellation of affective, cognitive, and somatic symptoms associated with functional abnormalities in relevant brain systems. Painful stimuli are primarily stressful and can trigger consistent responses in brain regions highly overlapping with the regions altered in MDD patients. Duloxetine has proven to be effective in treating both core emotional symptoms and somatic complaints in depression. This study aimed to assess the effects of duloxetine treatment on brain response to painful stimulation in MDD patients. A total of 13 patients and a reference group of 20 healthy subjects were assessed on three occasions (baseline, treatment week 1, and week 8) with functional magnetic resonance imaging (fMRI) during local application of painful heat stimulation. Treatment with duloxetine was associated with a significant reduction in brain responses to painful stimulation in MDD patients in regions generally showing abnormally enhanced activation at baseline. Clinical improvement was associated with pain-related activation reductions in the pregenual anterior cingulate cortex, right prefrontal cortex, and pons. Pontine changes were specifically related to clinical remission. Increased baseline activations in the right prefrontal cortex and reduced deactivations in the subgenual anterior cingulate cortex predicted treatment responders at week 8. This is the first fMRI study addressed to assess the effect of duloxetine in MDD. As a novel approach, the application of painful stimulation as a basic neural stressor proved to be effective in mapping brain response changes associated with antidepressant treatment and brain correlates of symptom improvement in regions of special relevance to MDD pathophysiology.