A 27 month prospective study concerned with aspects of patient-nurse assaults on the geriatric unit (three wards) of a Canadian provincial mental hospital was conducted. The findings were compared with those for the other words. The incidence of assaults was approximately the same in both areas (0.24-0.25 assaults/occupied bed/year respectively). In the former, the commonest diagnoses associated with assaultiveness were dementia followed by schizophrenia and in the rest of the hospital, schizophrenia. However, when base rates of assaultiveness were calculated allowing for the disproportionate number of patients with these conditions, mental retardation, and dementia were approximately twice as likely to be related to assaultiveness as schizophrenia; regardless of where the patients were located. In the geriatric unit attacks were more likely when patients were being physically guided or led and during the administration of drugs; elsewhere whilst physical restraints were being applied. In both hospital areas a comparatively small number of patients accounted for a disproportionate number of assaults and a few nurses were attacked repeatedly. The majority of episodes were trivial but in isolated cases personnel were off work for several months. The discussion focuses on the possibility of generalizing results.