The surprise question (SQ), "Would I be surprised if this patient died within one year?", is a simple instrument to identify patients with palliative care needs. The SQ-performance has not been evaluated in patients with advanced cancer visiting the emergency department (ED). To evaluate SQ's test characteristics and predictive value in patients with advanced cancer visiting the ED. Observational cohort study. Patients >18 years with advanced cancer in the palliative phase visiting the ED of an academic medical center. Attending physicians answered the SQ (not surprised [NS] or surprised [S]) and estimated Eastern Cooperative Oncology Group (ECOG)-performance status. Disease, visit, and follow-up characteristics were retrospectively collected from charts. SQ's sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV), and Harrell's c-index were calculated. Prognostic values of SQ and other variables were assessed by using Cox proportional hazards models. Two-hundred-and-forty-five patients were included (203 NS [83%] and 42 S [17%]), median age 62 years, 48% male. Follow-up on overall survival was updated until February 2019. At ED entry, NS-patients had worse ECOG-performance and more symptoms. At study closure, 233 patients had died (95%). Median survival was three months for NS-patients (interquartile [IQ]-range: 1-8); nine months for S-patients (IQ-range: 3-28) ( < 0.0001). SQ-performance for one-year mortality: sensitivity 89%, specificity 40%, PPV 85%, NPV 50%, c-index 0.56, and hazard ratio 2.1 for approaching death. ECOG 3-4 predicted death in NS-patients; addition to the SQ improved c-index (0.65); sensitivity (40%), specificity (92%), PPV (95%), and NPV (29%). At the ED, the SQ plus ECOG 3-4 helps identifying patients with advanced cancer and a limited life expectancy. Its use supports initiating appropriate care related to urgency of palliative care needs.