There remains much scientific, clinical, and ethical controversy concerning the use of electroconvulsive therapy (ECT) for psychiatric disorders stemming from a lack of information and knowledge about how such treatment might work, given its nonspecific and spatially unfocused nature. The mode of action of ECT has even been ascribed to a "barbaric" form of placebo effect. Here we show differential, highly specific, spatially distributed effects of ECT on regional brain structure in two populations: patients with unipolar or bipolar disorder. Unipolar and bipolar disorders respond differentially to ECT and the associated local brain-volume changes, which occur in areas previously associated with these diseases, correlate with symptom severity and the therapeutic effect. Our unique evidence shows that electrophysical therapeutic effects, although applied generally, take on regional significance through interactions with brain pathophysiology.magnetic resonance imaging | voxel-based morphometry | unipolar depression | hippocampus E lectroconvulsive therapy (ECT) is the oldest well-established procedure for somatic treatment of unipolar and bipolar disorders (1); however, its precise mechanism of action is still unclear (2). Our current understanding is that the antidepressant effect of ECT is partially mediated by seizure-induced neurotrophic effects, resulting in increased rates of neurogenesis, synaptogenesis, and glial proliferation, particularly in the hippocampus (3-5). Modern neuroimaging experiments have shown a critical role for the subgenual cortex (Brodmann area 25) (6) and deep brain stimulation of this area also results in alleviation of symptoms (7,8). Concerns regarding structural brain damage caused by ECT have been largely attenuated because of a lack of experimental evidence for ECT-induced neuronal damage (9). The scarce in vivo evidence for ECT-induced structural brain plasticity comes from region-of-interest (ROI) imaging studies reporting ECT-related hippocampal volume increases (10) that correlate with clinical outcome (11,12). Limited by an ROI approach and a lack of adequate control groups, these studies may have incompletely detected the effects of right unilateral ECT and failed to distinguish them from pharmacologically induced changes or indeed the effects of disease.We decided to resolve these ambiguities by carrying out a study with drug responsive (no-ECT) and drug-resistant (ECT) patients with either uni-or bipolar depression ( Fig. 1 and Table 1). The two psychiatric conditions are considered separate from pathophysiological and nosological viewpoints. A group of normal volunteers was included to control for non-ECT-associated confounds and to provide a way of assessing ECT-associated therapeutic effects. All subjects were recruited and imaged using structural magnetic resonance at entry (time point 1, TP1). If unresponsive to drugs, patients had ECT administered unilaterally to the right hemisphere. All study participants were imaged again at 3 mo (TP2) and 6 mo (TP3). ECT was giv...