We reviewed 32 studies to determine whether patient-related factors predict the effectiveness of cardiopulmonary resuscitation (CPR) of hospitalized patients. Limitations of individual studies are described. Single factors and multiple factors, as used in the Pre-arrest Morbidity (PAM) Index, were examined for their ability to predict CPR outcome. Although certain clinical conditions appear to predict CPR survival best, other individual patient-related factors influence the effectiveness of CPR. The PAM Index is superior to single factors in predicting survival after CPR in hospitalized patients, possibly because it is multifactorial. As some studies reveal efficacy of only some of the variables of the PAM Index, a modification of the PAM Index is proposed as an effective tool for predicting CPR outcome. Such a modification may offer geriatricians a practical basis on which some difficult 'do not resuscitate' decisions can be based.