Unwell hospital patients who are approaching the end of their lives and may die at any moment receive inconsistent care and often do not have opportunities to discuss their future care preferences. The AMBER (Assessment; Management; Best practice; Engagement; Recovery uncertain) care bundle was developed to help identify such patients, train health-care professionals to better communicate their concerns with them and their families and, where possible, to realise their preferences for place of care and death. The Liverpool Care Pathway for the Dying Patient, previously used across England, was designed to provide the best possible quality of care to those at the end of life. However, an independent review identified that it often was not used appropriately, leading to poor patient outcomes. A number of the criticisms of the Liverpool Care Pathway for the Dying Patient, such as reports of quickened deaths from the withdrawal of hydration/nutrition, as well as poor communication with patients and families, may have been identified earlier if it had been thoroughly evaluated. The AMBER care bundle, developed at Guy's and St Thomas' Hospital, aims to provide better outcomes for patients with clinically uncertain recovery and their families. It is important that the AMBER care bundle is properly investigated before wider use. A bigger study of the AMBER care bundle would be complex and expensive, so we examined whether or not this would be possible and acceptable to patients cared for in four wards across four hospitals, before deciding whether or not to go ahead with this bigger study. Two wards used the care bundle and two did not. We interviewed participants, their families and staff, and examined participants' clinical notes. We found that the AMBER care bundle was largely acceptable to patients, relatives and staff, and generally delivered as intended. We successfully collected information from 65 unwell patients at the beginning of the trial and again 3­5 and 10­15 days later. However, a limited number of data were collected at the final time point (10­15 days) due to many participants being discharged from the hospital. Group discussions with staff and interviews with participants and relatives identified important changes required to improve the AMBER care bundle and views on how the trial was conducted. These included simplifying the type of patients who may be appropriate for the AMBER care bundle and improving communication training for staff. Although we identified that a further study was technically possible, it is currently impractical. Future solutions that would require further testing include focusing on clinical need rather than trying to guess how the patient's condition will develop to identify potential trial participants and using questions completed by the patients as part of their routine care as a source of information. In the meantime, the AMBER care bundle continues to be used in over 40 hospitals in England.