One third of oncological treatment costs per patient is allocated to the last phase of life. In the era of molecular oncology and immuno-oncology, patients benefit from new treatment options inducing durable and long-lasting responses. However, it becomes more difficult to estimate the prognosis of oncology patients. The treatment indication is based on the evidence from randomized controlled studies. In contrast, the decision, when to stop treatment at the end of life and provide best supportive care, is an emerging and challenging situation in routine clinical care of oncologists and palliative care teams. Up to 50% of oncology patients receive chemotherapy within the last 4 weeks before death, thus it becomes evident to stop futile treatment. Reliable biomarkers to predict the response of immunotherapy are lacking for most of solid tumors. Several palliative prognostic scores have been validated to calculate the probability of survival in the next 30 - 60 days. Unfortunately, there is no consensus on which score should be preferred and none was validated in period of immuno-oncology. The estimation of expectation of life by an interdisciplinary medical team is recommended by the German guideline of palliative medicine. Of note, treating physicians often overestimate the prognosis of patients, and shared decision making whether to start, to continue or to stop therapy for the individual patient remains difficult. Early integration of palliative medicine and advance care planning focus on the patient's medical perspective. Clinical trials investigating the integration of oncology and palliative care point to health gains: improved survival and symptom control, less anxiety and depression, reduced use of futile chemotherapy at the end of life, improved family satisfaction and quality of life, and improved use of health-care resources. Changes at the system level are necessary for implementation of advance care planning to improve the quality of the end of life of oncology patients., Competing Interests: Erklärung zu finanziellen Interessen Forschungsförderung erhalten: nein; Honorar/geldwerten Vorteil für Referententätigkeit erhalten: ja, von einer anderen Institution (Pharma- oder Medizintechnikfirma usw.); Bezahlter Berater/interner Schulungsreferent/Gehaltsempfänger: ja, von einer anderen Institution (Pharma- oder Medizintechnikfirma usw.); Patent/Geschäftsanteile/Aktien (Autor/Partner, Ehepartner, Kinder) an Firma (Nicht-Sponsor der Veranstaltung): nein; Patent/Geschäftsanteile/Aktien (Autor/Partner, Ehepartner, Kinder) an Firma (Sponsor der Veranstaltung): nein. Erklärung zu nichtfinanziellen Interessen Die Autorinnen/Autoren geben an, dass kein Interessenkonflikt besteht., (Georg Thieme Verlag KG Stuttgart · New York.)