Aim One condition associated with severe end‐of‐life pain that can lead to a poor quality of death is cancer. Cancer pain in people with dementia is of particular concern because of communication problems that occur with worsening disease. The aim of the current pilot study was to examine the association between hospice enrolment, dementia severity and pain among nursing home residents who died from advanced cancer. Methods Between‐groups cross‐sectional chart audits of 55 nursing home residents with dementia who died from cancer were carried out. Results A total of 45% of residents were in hospice at the end‐of‐life. Residents in hospice were more likely to receive an opioid (80% vs 43%, P = 0.005); but less likely to show severe cognitive impairment (20% vs 50%, P = 0.050). Enrolment in hospice was associated with an increased likelihood of receiving an opioid after controlling for level of cognitive impairment (OR = 3.9, 95% CI = 1.1–14.0, P = 0.037). Lower levels of cognitive functioning were associated with a decreased likelihood of receiving an opioid after controlling for enrolment in hospice (OR = 0.3, 95% CI = 0.1–0.8, P = 0.030). Notably, 40% of nursing home residents with dementia who died from cancer did not receive any opioid during this time. Conclusions Preliminary results suggest that hospice enrolment might be influenced by the facility or region of this particular country. Hospice enrolment predicts more opioid pain treatment in residents with dementia and terminal cancer; however, no resident with very severe dementia and terminal cancer was placed in hospice care. Severely cognitively impaired nursing home residents requiring opioids are at great risk of suffering from untreated advanced cancer pain. New methods are urgently required to improve end‐of‐life palliative care for nursing home residents with terminal cancer and severe dementia. Geriatr Gerontol Int 2013; ●●: ●●–●●.