Identifying and quantifying the location of pain may be important for understanding specific functional impairments in elderly populations. The purpose of the present analysis was two-fold: first, to describe the reliability of a scoring method for the McGill Pain Map (MPM), and second, to validate the method of scoring the MPM as a tool for assessing areas of body pain in an epidemiologic study. In interviews performed at the subjects' homes, 411 community dwelling Mexican-American and non-Hispanic white subjects aged 65-74 from the San Antonio Longitudinal Study of Aging (SALSA) were asked to describe the location of their pain on the map of the human body included in the McGill Pain Questionnaire. The location of pain was scored by overlaying the survey figures with a MPM template divided into 36 anatomical areas. Inter- and intra-rater agreement among three raters was measured by calculating a kappa statistic for each of the body areas, and an intraclass correlation coefficient for the total number of painful areas (NPA). Internal validity was measured by Spearman's rho between the NPA and the Present Pain Index (PPI) and Pain Rating Index (PRI) of the McGill Pain Questionnaire, and external validity by correlation between NPA and the Perceived Health (PH), Amount of Bodily Pain (APB), and Pain Interference with Work (PIW) items of the Medical Outcomes Study, and the Perceived Physical Health (PPH) question of the San Antonio Heart Study. Average inter-rater agreement for individual MPM areas was 0.92 +/- 0.01, and average agreement for NPA was 0.96 +/- 0.01. Intra-rater agreement for individual areas averaged 0.94 +/- 0.01, and for NPA = 0.99 +/- 0.001. Pain in one or more areas was present in 47.7% of the subjects. For the whole sample, correlations between NPA and the validation indices were: PPI (0.91), PRI (0.89), PH (0.25), ABP (0.64), PIW (0.49), and PPH (0.20). Among the 196 subjects with pain, correlations were: PPI (0.34), PRI (0.34), PH (0.19), ABP (0.21), PIW (0.38), and PPH (0.19)-p < 0.01 for all correlations. In conclusion, we have developed a reliable method of scoring the MPM and have shown evidence of its validity in a community-based sample of elderly subjects. Patterns of painful body areas may be associated with specific diseases and functional impairments.