The cerebral structure and function of a group of ten men admitted to a special (high~security) hospital in the south of England were assessed using a range of methods. All had committed at least one act of very serious violence, or similarly dangerous acts (e.g. arson). Six of the group had schizophrenia and four a primary personality disorder. The men were all regarded as still posing a sufficiently high risk to others that they would not have been allowed to leave the hospital without a multi~ ple nurse escort. Previous reports suggest that magnetic resonance imaging (MRI) is better than computed tomography (CT) in detecting structural cerebral lesions in people with a mental disorder. The main purpose of this study, therefore, was to test the feasibility of using an MRI scanner which could be taken into the hospital to eval~ uate the men, and then undertake a preliminary review of the relative roles of the dif ferent approaches to cerebral investigation. All the men tolerated the MRI without incident. Results from neuropsychological testing and MRI were similar insofar as both indicated definite cerebral pathology in nine of the ten cases, while the CT scan suggested damage in just half of the cases and the EEG in only six men. In six cases MRI demonstrated atrophy of mesial temporal structures that had not been detected with CT. In only one case did the CT scan provide evidence of localised disturbance, and the EEG was only slightly better in this regard. There were discrepancies in localisation of damage between the neuropsychological testing and the MRI results. A full~scale, systematic study of mentally disordered serious offenders using a broad~ based approach seems important. Thus far this is the largest such series.