Consensus exists that a do-not-attempt-resuscitation order (DNAR) is appropriate if a resuscitation attempt is futile. Less agreement exists when this point is reached. We investigated the influence of three major considerations for in-hospital DNAR orders: expected survival probability after resuscitation, prospects of the patients' current condition without a cardiac arrest and the patients' autonomous decision not to want resuscitation. We calculated an expected survival probability according to two multi-morbidity prediction scores for each patient, assuming the event of cardiac arrest. The prospects of the current condition without a cardiac arrest was estimated by the patients' physician, in terms of life expectancy and quality of life (level of dependency after discharge and pain). The patients' preference was documented from the medical records. A total of 470 patients were included in the study. Fifty-eight patients (12%) had a DNAR-order, 11 of these patients (19%) wanted no resuscitation. The patients' prospects (life expectancy, dependency after discharge), and age proved to be independently associated with the presence of a DNAR order. The odds ratio (OR) for the presence of a DNAR order was 37 (CL 14-107) for an estimated life expectancy less than 3 months, 13 (CL 4-41) for a life in a nursing home and four (CL 2-12) for an age of 80 years and older. Expected survival probability after resuscitation and pain were not independently associated with a DNAR order. We conclude that resuscitation is considered futile on the basis of the patients' age and prospects without cardiac arrest and that the impact of expected survival probability on these decisions is small.