The recent advances in the miniaturisation of electronic activity monitoring devices have opened up new methods for the study of human activity. This thesis describes the application of such technologies to the measurement of behavioural disturbance in dementia. Many behavioural problems in dementia are manifested as changes in levels or patterns of activity. However, the validity and reliability of the use of activity monitors in the study of dementia had not been systematically addressed. The first study carried out established that 80% of severe dementia hospital in-patients (22 subjects) would tolerate wearing an activity monitor for 72 hours. Evening activity levels in these patients were found to correlate with nurse ratings of 'wandering and pacing' behaviour. Two healthy elderly control samples were examined for activity level differences between seventy year olds (33 subjects: 70-72 years) and eighty year olds (25 subjects: 80-84 years). No significant differences were observed. A comparison of the severe dementia patients and controls showed significantly less activity in the dementia patients with a rise towards late afternoon whereas the control subjects showed peak activity in the morning. The validity of monitoring activity as a measure of behavioural disturbance was assessed using a sample of mixed severity dementia patients living in the community (25 subjects). Cognitive function was assessed with the Mini-Mental State Examination and the Cognitive Assessment Schedule. Behavioural ratings were assessed using the Revised Memory and Behaviour Problem Check-list and the Behaviour Rating Scale of the Clifton Assessment Procedures for the Elderly. Collective ratings of behavioural problems were found to correlate highly (> 0.5) with cognitive impairment. A rating of general agitation showed a significant correlation with afternoon activity measures, although other ratings of behavioural problems did not correlate significantly with activity.