Repetitive transcranial magnetic stimulation (rTMS) is being investigated as an alternative treatment for depression. However, little is known about the clinical role and the neurophysiological mechanisms of the action of rTMS in these patients. In this study, 99mTc-HMPAO single photon emission computed tomography (SPECT) was used to map the effects of left dorsolateral prefrontal rTMS on prefrontal activity in seven patients who met DSM-IV criteria for major depression resistant to pharmacological treatment. rTMS consisted of 30 trains of 2-s duration stimuli (20 Hz, 90% of motor threshold), separated by 30-s pauses. Each patient underwent three SPECTs: at baseline; during the first rTMS; and 1 week after 10 daily sessions of rTMS. Regional cerebral blood flow (rCBF) of each cerebral region was normalized to the rCBF value in the cerebellum and relative changes in normalized rCBF were addressed using a region-of-interest analysis. The Hamilton Depression Rating Scale (HDRS) was used for clinical evaluation before and after rTMS. A significant rCBF increase after the 10 sessions of rTMS was found in the left prefrontal region (MANOVA F=5.29, d.f.=2,10, P=0.027), but no significant rCBF changes were found during the first rTMS session. The remaining cerebral regions showed no significant rCBF changes at any time. Only two patients showed a clinical improvement after rTMS, with 50% reduction of the initial HDRS score. The study was repeated under placebo conditions (identical design but addressing coil discharges to the air) in these two patients, who failed to show any rCBF increase during sham-rTMS. No relationship was found between the percentage of left prefrontal rCBF change and the clinical findings. In conclusion, rTMS of the left prefrontal cortex induces a significant rCBF increase in this region, despite the limited clinical effect in our sample of depressed patients. Cerebral perfusion SPECT is a useful tool to map cerebral activity changes induced by rTMS.