-Cardiopulmonary resuscitation (CPR) is attempted on too many patients. At its best, CPR is the gift of life: chest compression, ventilation, intravenous medication and defibrillation followed by years of productive and fulfilled being. At its worst [3][4][5] , it offers a scenario of vomit, blood and urine, then a confused, brain-damaged twilight, breathlessness from a failing ventricle, pain from rib fractures, until expiring in thrall to the full panoply of the intensive care unit or forgotten in the long darkness of the persistent vegetative state. No humane doctor would consider this a good death, nor would any poet, priest, painter, musician or novelist use images of CPR to represent the Good Death. Rather, the images are more likely to be those of the factory: death in the industrial age. CPR is a lottery -the odds are good for some and the winnings are life itself. For most, the odds are poor and the outcome for some worse than the death that CPR is designed to avoid. The key to better outcome is better selection, but we often do not know how to select, even if we do know when to stop 6 . An effective treatment for the select few becomes an industrial rite of passage for the many. Our problem is how does the competent patient -or the doctor trying to fathom the incompetent patient's best interests -determine whether to buy the lottery ticket? In few areas of medicine is that truer than in the debates around CPR. Surveys abound, but the practical guidance given to clinicians is limited. In one study, 18% thought CPR should be given to patients with 'terminal cancer' 8 ; in another, six patients with carcinomatosis wanted CPR 9 . Hill et al 10 reported that one-third of doctors would resuscitate patients with incurable malignancy, yet concordance between doctors' and patients' views are little better than chance. In a study of 1,438 seriously ill patients, health values were found to vary enormously, changed over time, related to few other preferences or health status measures and were higher than surrogates thought 11 . Despite the belief that they knew patients well, physicians were no better than interns in assessing the preferences for CPR of 230 seriously ill adults and both had only a fair understanding of those preferences or quality of life 12 .It was against that background that last April the charity Age Concern mounted a campaign over the alleged practice of designating elderly patients as 'not for resuscitation' . The index case was a 67-year-old woman, Jill Baker, suffering from cancer. She was reported as having 'spotted a "do not resuscitate" (DNR) order on her medical notes' after she was admitted for treatment at St Mary's Hospital, Portsmouth 13 . The reported details of Mrs Baker's case may suggest that the DNR decision had more to do with her incurable malignancy and sepsis than her age, but the implications of this possibility -equally significant and controversial -went unexplored. Age Concern went on to claim that it knew of more than 100 similar cases that it promised to br...