Despite its high prevalence, surprisingly little is known regarding the longitudinal course of activity limiting pain in community dwelling elders. This paper uses data from a 6 year study of 600 individuals age 70 and older who resided in Philadelphia to address this issue. In-home health interviews, equally divided between those of Caucasian or African American descent, were conducted at study start and repeated 2, 4, and 6 years later. Questions taken from the McGill Pain Scale were used to classify individuals as having: no pain, pain without activity limitations, or pain with activity limitations. Based on 319 participants who had 2 or 3 follow-up interviews, 5 longitudinal pain level patterns were identified: No pain, 24.5% of respondents; Pain without activity limits (10.3%); Pain with activity limits (27.6%); New Pain or pain that worsened (24.8%); and Pain that diminished (12.9%). Multinomial logistic regression analyses revealed that, compared to the “No pain” group, the odds of being in the “Pain without activity limits” group or the “Pain with activity limits” group increased with the number of health conditions and depressive symptoms reported during the first interview. Neither age nor race changed the odds of reporting chronic pain with or without activity limitations. Women were 5 times more likely to report chronic pain without activity limits and 2.5 times more likely to report chronic activity limiting pain. While these findings are consistent with those from cross-sectional studies, they demonstrate the remarkable stability of pain experienced by older individuals over time and suggest that successful management of chronic pain in the elderly, particularly women, challenges practitioners, educators and health planners.