Postmortem magnetic resonance (MR) scans were performed on the brains of 12 victims of fatal head injuries. These were compared with neuropathological studies of the entire brain. The first six subjects were imaged with the brain in situ and comparison was also made with antemortem computed tomography (CT). The brains from the subsequent six subjects were removed at autopsy, fixed in formalin and then imaged in a mitre box designed to overcome the problems encountered in the pilot study. Although both CT and MR imaging (MRI) detected all clinically relevant haemorrhagic lesions, many pathologically significant lesions were missed. MRI detected many more lesions than CT, but still failed to visualize areas of non-haemorrhagic axonal injury.