A retrospective study of a consecutive set of patients (n=264) undergoing CPR revealed that pre-arrest factors measured by three different morbidity scores can accurately predict a subset of patients for whom CPR attempts are futile. These morbidity scores have low sensitivity but high specificity. Whilst there is widespread acceptance that the patient's views about CPR should be sought there is little evidence about patient's opinions. Three different populations of elderly patients (mixed medical inpatients n=214, patients in the community following an admission with stroke n=100 and inpatients with acute stroke n=60) were interviewed to determine which life-sustaining treatments they would accept/refuse. Inclusion rates for the hospital-based studies were low (36%, 75%) because many inpatients were physically or mentally unable to participate in decisions about CPR. Patients had poor knowledge of techniques and success rates of CPR and following education around 10% refused CPR who had previously accepted it. The acceptance rate for CPR in the three groups was 60% to 82% for CPR, and 60% to 72% for artificial feeding. A significant number of patients had changed their minds about CPR between admission and follow-up at discharge from hospital (16%), or in community (17%). Although competency rates were higher at follow-up this factor alone could not account for changes in view about CPR. There was no significant relationship found between age, sex, disability or quality of life and wish for CPR. Eight qualitative interviews were carried out which allowed a more detailed investigation of the influences and reasoning behind CPR decisions. Advance directives are advocated as a means of improving patient involvement in CPR decisions. A questionnaire revealed that General Practitioners are often ignorant of their legal status. Only 70% of inpatients questioned following stroke were competent to reasons why they were reluctant to complete advance directives.