BACKGROUND: The Palliative Medicine Unit at University Hospital of Trondheim, Norway, started an intervention programme that aims to enable patients to spend more time at home and die there if they prefer. Close cooperation was needed with the community health-care professionals, who acted as the principal formal caregivers, and a multidisciplinary consultant team coordinated the care. We did a cluster randomised trial to assess the intervention's effectiveness compared with conventional care METHODS: Community health-care districts in and around Trondheim, Norway, were defined as the clusters to be randomised. We enrolled 434 patients (235 assigned intervention and 199 conventional care [controls]) in these districts who had incurable malignant disease and an expected survival of 2-9 months. Main outcomes were place of death and time spent in institutions in the last month of life. FINDINGS: 395 patients died. Of these, more intervention patients than controls died at home (54 [25%] vs 26 [15%], p<0.05). The time spent at home was not significantly increased, although intervention patients spent a smaller proportion of time in nursing homes in the last month of life than did controls (7.2 vs 14.6%, p<0.05). Hospital use was similar in the two groups. INTERPRETATION: The palliative-care intervention enabled more patients to die at home. More resources for care in the home (palliative care training and staff) and an increased focus on use of nursing homes would be necessary, however, to increase time at home and reduce hospital admissions.