Background: Documented resuscitation orders have relevance in the management of a pulseless, unresponsive patient. Although useful, the frequency of their documentation in the case notes of newly admitted medical patients is not well established. Aim: To investigate the frequency of early clear documentation of resuscitation orders in patients' admission notes. Design: Retrospective audit. Methods: The admission notes of 618 medical admissions to an Australian tertiary referral teaching hospital between January and December 2007 were reviewed to calculate the frequency of clear resuscitation documentation. Certain outcomes of each admission, such as in-hospital death, were obtained via hospital-based computerized records. Results: Within the first 24 h of admission, discussions regarding resuscitation were not documented for 78% of patients. Of the 482 patients with no documented resuscitation orders, 5 patients died during their index admission. Of the 136 patients with documented resuscitation orders, 24 patients died during their index admission. As age or a measure of clinical debility increased, the absolute number and relative proportion of resuscitation discussions increased significantly (P<0.0001) and the number and proportion of patients deemed not for resuscitation also increased (P<0.0001). Conclusions: Those patients apparently targeted for discussion were older, more frail and acutely unwell. We propose widespread use of a clinical scoring system to identify those patients who need their resuscitation status clarified early in their admission prior to clinical deterioration.