Purpose: We developed and evaluated an explicit procedure for obtaining self-report pain data from nursing home residents across a broad range of cognitive status, and we evaluated the consistency, stability, and concurrent validity of resident responses. Design and Methods: Using a modification of the Geriatric Pain Measure (GPM-M2), we interviewed 61 residents from two nursing homes (Mini-Mental State Examination score, M = 15 plus or minus 7) once a week for 4 consecutive weeks. We collected additional data by means of chart review, cognitive status assessments, and surveys of certified nursing assistants. We used descriptive and correlational analyses to address our primary aims. Results: Eighty-nine percent of residents completed all four scheduled interviews. Cognitive status was not significantly correlated with number of nonresponses and prompts for yes-no questions, but it was significantly correlated with nonresponses and prompts for Likert-scale questions (r = -0.48, p less than 0.001 and r = -0.59, p less than 0.001, respectively). Completion time for the 17-item pain measure (M = 13 min) was not predicted by cognitive status. Residents' scores on the GPM-M2 were significantly correlated with number of chronic pain-associated diagnoses, r = 0.37, p less than 0.01, and internal consistency was excellent, alpha = 0.87 - 0.91. Residents' GPM-M2 scores were stable over time, r =0.74 - 0.80, p less than 0.0001, for all comparisons. Implications: Using explicit protocols and reporting procedural data allows researchers and clinicians to better understand and apply results of self-report studies with cognitively impaired elders. Results suggest that many nursing home residents can provide consistent and reliable self-report pain data, given appropriate time and assistance. (Contains 3 tables.)