Background: In the last decade, the use of sedative substances to keep a patient in deep sedation until death increased fourfold in German-speaking Switzerland, where every third patient admitted to hospital, palliative care unit or hospice died continuously deeply sedated. Aim: To investigate sedation practices across healthcare settings and to identify their associations with conventional symptom control. Design: National mortality follow-back study in Switzerland between 2013 and 2014. Questionnaires on medical end-of-life decisions were sent to attending physicians of a continuous random sample of all registered deaths aged 1 year or older. Setting/participants: Of all sampled deaths, 3678 individuals who died non-suddenly and not through an external cause were included. Results: Across settings, continuous deep sedation appeared more likely in patients aged younger than 65 years (odds ratio range: 1.53–2.34) and as part of or after intensified alleviation of pain and symptoms (odds ratio range: 1.90–10.27). In hospitals, sedation was less likely for cancer patients (odds ratio: 0.7, 95% confidence interval: 0.5–1.0, p = 0.022). In nursing homes, sedation was more likely for people who were married (odds ratio: 1.8, 95% confidence interval 1.3–2.5, p = 0.001). Conclusion: In all settings, sedated patients have significantly more pain problems compared to patients not receiving sedation. Large differences between settings seem to indicate different patient populations, different levels of professionals’ palliative care experience and different availability of treatment options. Our study suggests that certain patient groups who may be as vulnerable to refractory pain and symptoms as others are less likely to receive continuous deep sedation until death when warranted.