Ameliorating pain symptoms in persons with severe dementia is difficult due to the inherent problems of pain assessment and communication with this population. This study aims to examine the utility of nine pain assessments in identifying pain and documenting responsiveness to pain medication in 121 nursing home residents with dementia. The assessments included self-report, informant ratings by nursing staff, and direct observations. This was an open medication study, with comparison groups of persons with dementia. Those evaluated as having pain were included in the pain medication protocol. Three comparison groups were utilized: those not considered in pain at baseline, those in pain at baseline for whom caregivers refused to follow the pain medication protocol, and a subset of those not in pain at baseline who were followed for a period of time similar in length to the treatment group. Informant ratings were the assessments most likely to detect pain and direct observations were least likely to do so. Both self-report and informant ratings showed a significantly greater decrease in pain in the treatment group relative to the comparison groups. All participants who followed the pain protocol reached a pain free stage by the study criteria. A limitation of the study was the reduced endpoint number due to family or physician resistance to following treatment recommendations. The results present a methodology that caregivers can use to increase the detection of pain in persons with dementia, enabling them to treat it more effectively, and thereby improve life quality for dementia patients.