Background: Aggression in psychoses is of high clinical importance, and volumetric MRI techniques have been used to explore its structural brain correlates.Methods: We conducted a systematic review searching EMBASE, ScienceDirect, and PsycINFO through September 2017 using thesauri representing aggression, psychosis, and brain imaging. We calculated effect sizes for each study and mean Hedge's g for whole brain (WB) volume. Methodological quality was established using the PRISMA checklist (PROSPERO: CRD42014014461).Results: Our sample consisted of 12 studies with 470 patients and 155 healthy controls (HC). After subtracting subjects due to cohort overlaps, 314 patients and 96 HC remained. Qualitative analyses showed lower volumes of WB, prefrontal regions, temporal lobe, hippocampus, thalamus and cerebellum, and higher volumes of lateral ventricles, amygdala, and putamen in violent vs. non-violent people with schizophrenia. In quantitative analyses, violent persons with schizophrenia exhibited a significantly lower WB volume than HC (p = 0.004), and also lower than non-violent persons with schizophrenia (p = 0.007).Conclusions: We reviewed evidence for differences in brain volume correlates of aggression in persons with schizophrenia. Our results point toward a reduced whole brain volume in violent as opposed to non-violent persons with schizophrenia. However, considerable sample overlap in the literature, lack of reporting of potential confounding variables, and missing research on affective psychoses limit our explanatory power. To permit stronger conclusions, further studies evaluating structural correlates of aggression in psychotic disorders are needed.