Dignity of the natural end of life for everybody is one of the new great challenges of medicine and social care for the beginning 21st century. However, many end of life care providing doctors are confused about how to categorize the help they give. One of the central problems is predicting the life expectancy of an individual patient. Difficulties in this field can become ethical dilemmas when physicians are obliged to predict accurately a patient's prognosis as the basis for a certain care strategy. Clinical estimation of the duration of life for patients with end of life cancer needs experience and training. Education programmes in the field should include this topic much more until now. Prognosis should be based more on proven indices and less on intuition. However, there is no doubt that daily clinical practice limits the use of highly sophisticated computer-based score models. Even maximal accuracy of prognosis will not exclude the risk of errors in a great part of patients. This limits their classification in care categories too strictly defined. Health care systems should avoid models for care with standards and budgets based on prognostic estimates and the medical community should avoid claim by disciplines of certain categories of patients defined by their prognoses. What we need is a network of assistance for incurable patients with single parts defined by patients needs and not by predicted life expectancy. Separating palliative and terminal care is artificial and often in contrast to the needs of the patients.