The Edmonton Regional Palliative Care Program was established to increase access to palliative care for terminal cancer patients in the region. Inpatient care is delivered, in decreasing order to distress, at the tertiary palliative care unit, by consult teams in acute care facilities, and in hospices. We reviewed the admission data for all patients discharged from the program between November 1, 1997, and October 31, 1998, in order to determine if demographical and clinical variables suggested appropriate use of the three levels of care. Patients admitted to the tertiary palliative care unit were significantly younger than those admitted to acute care of hospices (61 years versus 68 years and 71 years respectively, p < 0.0001), had a higher frequency of positive screening for alcoholism in the CAGE questionnaire (27% versus 16% and 14% respectively, p < 0.0001), and a higher frequency of poor prognostic pain syndromes as compared to acute care admissions (87% versus 65%, p < 0.0001). Overall, frequency of symptoms and severe symptoms was significantly higher in patients admitted to the palliative care unit than those admitted to the other two settings. Our results suggest that patients with demographic and clinical indications of higher distress are more frequently admitted to the tertiary palliative care unit. The clinical tools are useful predictors of utilization that can be used for monitoring health care delivery.