Inconsistent results in studies of cost-effectiveness of home health care have led to the need for identification of target populations for whom cost-savings can be anticipated if expanded home care programs are introduced. This analysis of results of a randomized controlled study of efficacy of a physician/geriatric nurse practitioner/social worker home care team identifies such a potential target population. The team provides round-the-clock on-call medical services in the home when needed, in addition to usual nursing and other home care services, to home-bound chronically or terminally ill elderly patients. Overall health services utilization and estimated costs were not substantially different for the patients who did not die while in the study; however, for those who did die, team patients had considerably lower rates of hospitalization and overall cost than controls, and more frequently died at home. Of 21 team and 12 control patients who died but had at least two weeks of utilization experience in the study, team patients had about half the number of hospital days compared with controls during the terminal two weeks, and although they had more home care services, had only 69 per cent of the estimated total health care costs of the controls. Satisfaction with care received was significantly greater among the total group of team patients, and especially among their family caretakers, than among controls. This model is effective in providing appropriate medical care for seriously ill and terminal patients, and in enabling them to die at home if they so wish, while at the same time reducing costs of care during the terminal period.