Since an increasing number of cancer patients is surviving for a long time, there is also more attention for prevention of late undesirable effects of treatments, for example, by applying treatment de-escalation strategies. Unfortunately, it is often not possible to investigate de-escalation strategies in randomised studies. As long as there is no hard evidence for the effectiveness of de-escalation strategies, physicians have a joint responsibility: (a) to apply shared decision making when choosing the best suited treatment; (b) to prospectively collect and evaluate data on hard and soft outcome measures; and (c) to use research results for improving the information needed for shared decision making. Most patients with cancer have explicit choices to make nowadays. In order to make it possible to conduct successful randomised studies into de-escalation strategies in the future, a number of barriers will need to be removed. In addition, practitioners should organise care in ways that make it possible to learn from every patient, in order to develop a stronger basis for decision-making for patients and physicians in the future.