Five palliative home care teams participated in a prospective Swedish study that included 221 palliative cancer patients. All patients with incurable malignant disease that were admitted and died during 1999 were included. On admission, demographic data were recorded. When patients, despite ongoing home care, were referred to institutional care, doctors and nurses involved were interviewed about the reasons for this. After the patients' death next of kin involved in the care were interviewed according to a questionnaire. Approximately half of the patients died at home. The reasons for referral showed a wide diversity and included both social and psychosocial factors, medical emergencies and problems related to symptom control. A preference for dying at home and not living alone were shown to be the strongest predictors of home death (p = 0.001). However, 35% of patients living alone died at home. Interestingly enough, Karnofsky performance index (KPI) at admission was significantly lower for those dying at home, despite similar mean time of care. The understanding of impending death was significantly more common among the families of those patients dying at home.