In April 2000 Age Concern England publicized an account of a do-not-resuscitate (DNR) decision written in the hospital notes of an older patient without her knowledge, and the British media claimed doctors 'let older patients die' who could have been resuscitated. This paper explores this belief in the context of an ethnographic study of communication between medical and nursing staff on two acute surgical wards, which took place shortly before this media outcry. DNR decisions emerged as a key area in which staff felt their communication with each other, with patients and (to a lesser extent) with relatives was poor. Contrary to media accounts that 'too many' DNR decisions occurred, the need to consider withholding resuscitation was generally neglected by doctors in these wards, in contravention of British Medical Association/Royal College of Nursing guidelines. Nurses and junior doctors thought 'too few' DNR decisions were taken and sometimes 'too late' to prevent serious harm to dying patients subjected to futile resuscitation attempts. Evidence of ageism in DNR decision making was weak, but reflections are offered on the influence of 'social worth' and 'moral imperatives' in DNR decision making. The process of unearthing unexpected findings during ethnographic research and testing emerging themes opportunistically is briefly considered.