OBJECTIVE--To assess the preference of terminally ill patients with cancer for their place of final care. DESIGN--Prospective study of randomly selected patients with cancer from hospital and the community who were expected to die within a year. Patients expected to live less than two months were interviewed at two week intervals; otherwise patients were interviewed monthly. Their main carer was interviewed three months after the patient's death. SETTING--District general hospital, hospices, and patients' homes. MAIN OUTCOME MEASURE--Stated preferred place of final care; actual place of death; reason for final hospital admission for those in hospital; community care provision required for home care. RESULTS--Of 98 patients approached, 84 (86%) agreed to be interviewed, of whom 70 (83%) died during the study and 59 (84%) stated a preferred place of final care: 34 (58%) wished to die at home given existing circumstances, 12 (20%) in hospital, 12 (20%) in a hospice, and one (2%) elsewhere. Their own home was the preferred place of care for 17 (94%) of the patients who died there, whereas of the 32 patients who died in hospital 22 (69%) had stated a preference to die elsewhere. Had circumstances been more favourable 67% (41) of patients would have preferred to die at home, 16% (10) in hospital, and 15% (9) in hospice. CONCLUSION--With a limited increase in community care 50% more patients with cancer could be supported to die at home, as they and their carers would prefer.