Previous studies of critical care admissions have largely compared patients that have been granted or declined admission. To better understand the decision process itself, our ethnographic approach combined observation of and interviews with critical care physicians in a large English hospital. We observed 30 critical care doctors managing 71 referrals and conducted ten interviews with senior decision-makers to explore the themes raised by our observations. We analysed data using the constant comparative method. We found that the decision to move a patient to critical care was just one way in which the trajectory of critical illness could be modified. When patients were admitted to critical care, it was not always for invasive monitoring or advanced organ support, with some admitted for more general medical and/or nursing care. When patients were declined admission, they were not simply forgotten or left behind; they nevertheless underwent careful assessment and follow-up. Thus, depicting admission or refusal as a binary event is misleading. We suggest that prescriptive admission algorithms are problematic for clinicians, in that they may not take into account the complexity of clinical practice.