This study was prompted by public and professional concern that the use of opioids for symptom control might shorten life. We retrospectively analysed the pattern of opioid use in the last week of life in 238 consecutive patients who died in a palliative care unit. Median doses of opioid were low (26.4 mg) in the last 24 h of life and patients who received opioid increases at the end of life did not show shorter survival than those who received no increases. The doctrine of double effect therefore need not be invoked to provide symptom control at the end of life.