Background: Community-based palliative care may potentially benefit patients by offering their preferred care at the end of life and benefit systems by reducing hospital use. Objective: To compare place of death and acute care hospital use in the last year of life between cancer decedents who did and did not access a community-based palliative care service (PCS). Design: Retrospective observational cohort study using linked individual administrative records from cancer registry, hospital, emergency department (ED), mortality, and PCS databases. Propensity score-weighted regression methods were used. Setting/Subjects: Whole of population study incorporating 28,561 West Australian cancer decedents from 2001 to 2011. Measurements: Exposure was defined as ever/never accessed PCS. Outcomes were place of death (in/out of hospital) and the number, length of stay, and cumulative cost of hospital admissions at the end of life. Results: Decedents who accessed the service (n = 16,530) had triple (adjusted odds ratio 3.19 [3.01-3.38]) the odds of dying out of hospital compared with those who did not. Unplanned hospitalizations were reduced in the last year (adjusted incidence rate ratio [IRR] 0.94 [0.91-0.97]) and last week of life (adjusted [IRR] 0.35 [0.33-0.38]), as were ED presentations (adjusted RR 0.92 [0.98-0.95], adjusted RR 0.26 [0.23-0.28]) in the last year and last week of life, respectively. There were significant reductions in average total bed days (-7.60 [-8.34 to -6.87]) and acute care costs (-A$5,491 [-A$6,155 to -A$4,827]) over the last year of life. Conclusions: In addition to supporting people to die out of hospital, PCS was associated with reduced acute care admissions, bed days, and costs over the last year of life. The provision of high-quality palliative care in the community alleviates the burden on acute care hospitals and, thus, may partially offset public funding of this model.