| - The most widely accepted model of obsessive-compulsive disorder (OCD) assumes brain abnormalities in the "affective circuit", mainly consisting of volume reduction in the medial orbitofrontal, anterior cingulate and temporolimbic cortices, and tissue expansion in the striatum and thalamus. The advent of whole-brain, voxel-based morphometry (VBM) has provided increasing evidence that regions outside the "affective" orbitofronto-striatal circuit are involved in OCD. Nevertheless, potential confounds from the different image analysis methods, as well as other factors, such as patients' medication and comorbidity status, may limit generalization of results. In the present paper, we systematically reviewed the whole-brain VBM literature on OCD by focussing specifically on degree of consistency between studies, extent to which findings have been replicated and interrelation between clinical variables and OCD anatomy, a potentially crucial factor that has been systematically examined only in a limited number of studies. The PubMed database was searched through February 2012. A total of 156 studies were identified; 18 of them fulfilled the inclusion/exclusion criteria and included 511 patients and 504 controls. Results support the notion that the brain alterations responsible for OCD are represented at the network level, and that widespread structural abnormalities may contribute to neurobiological vulnerability to OCD. Apart from defects in regions within the classic "affective" circuit, volume reduction of the cortical source of the dorsolateral (DL) prefronto-striatal "executive" circuit (dorsomedial, DL, ventrolateral and frontopolar prefrontal cortices), and of reciprocally connected regions (temporo-parieto-occipital associative areas) is consistently described in OCD patients. Moreover, increased volume of the internal capsule and reduced frontal and parietal white matter volumes may account for altered anatomical connectivity in fronto-subcortical circuitry. Morphometric changes in both "affective" and "executive" parallel the disease clinical course, being at the same time responsible for variation in symptom severity. Thus, OCD mechanisms involve a more widespread network of cerebral dysfunctions than previously thought, which may explain the heterogeneity in clinical manifestations and symptom severity.