Background: Palliative sedation (ps), the continuous use of sedating doses of medication to intentionally reduce consciousness and relieve refractory symptoms at end of life, is ethically acceptable if administered according to standards of best practice. Procedural guidelines outlining the appropriate use of ps and the need for rigorous documentation have been developed. As a quality improvement strategy, we audited the practice and documentation of ps on our palliative care unit (pcu). Methods: A pharmacy database search of admissions in 2008 identified, for a subsequent chart review, patients who had received either a continuous infusion of midazolam (≥10 mg/24 h), regular parenteral dosing of methotrimeprazine (≥75 mg daily), or regular phenobarbital. Documentation of the decision-making process, consent, and medication use was collected using a data extraction form based on current international ps standards. Results: Interpretation and comparison of data were difficult because of an apparent lack of a consistent operational definition of ps. Patient records had no specific documentation in relation to ps initiation, to clearly identified refractory symptoms, and to informed consent in 60 (64.5%), 43 (46.2%), and 38 (40.9%) charts respectively. Variation in the medications used was marked: 54 patients (58%) were started on a single agent and 39 (42%), on multiple agents. The 40 patients (43%) started on midazolam alone received a mean daily dose of 21.4 mg (standard deviation: 24.6 mg). Conclusions: The lack of documentation and standardized practice of ps on our pcu has resulted in a quality improvement program to address those gaps. They also highlight the importance of conducting research and developing clinical guidelines in this area.