The aim of this study is to determine the accuracy, feasibility and acceptability of the surprise question (SQ) in combination with a clinical prediction tool (Supportive and Palliative Care Indicator Tool (SPICT)) in identifying residents who have palliative care needs in residential aged care facilities (RACFs) in Australia. A prospective cohort study in two RACFs containing both high-level care (including dementia) and low-level care beds. Directors of Nursing screened 187 residents at risk of dying by 12 months using first the SQ, and if positive, then the SPICT. At 12-months follow-up, deaths, hospitalisations, use of palliative care services, end-of-life care and clinical indicators were recorded. The SQ had a sensitivity of 70%, a specificity of 69.6%, a positive predictive value of 40.6% and a negative predictive value of 88.7% for death. All residents identified by the SQ had at least two general indicators of deterioration, while 98.8% had at least one disease-specific indicator on the SPICT. The SPICT marginally increased the ability to identify residents in need of proactive end-of-life planning. A combination of the SQ and the SPICT is effective in predicting palliative care needs in residents of aged care facilities, and may trigger timely care planning.