The proposed research seeks to understand the process by which an older adult who is experiencing an episode of illness decides whether to seek the care of a physician. Two complementary methodological approaches are taken. (l) Detailed illness episode data from a seven wave longitudinal panel study of 1009 Medicare recipients enrolled in a Health Maintenance Organization (HMO) will be used to develop a series of event history models of physician contact. The effects of various explanatory variables (sociodemographic, prior health history) and time.varying covariates (other illness response strategies) on the risk of seeking medical care will be estimated in proportional hazards models that control for specific illness types and respondent categories. (2) A sample of 150 ethnographic informants will be selected from the original group of respondents to represent theoretically relevant categories of the population (e.g., men, women, married, not married, U.S. born, immigrants, health-status). Detailed ethnographic data on illnesses experienced in later life, their causes and symptoms, and the range of available and appropriate treatment alternatives for each illness will be collected systematically using free lists, card sorts, paired comparisons, and sentence frames. These data will be analyzed using techniques such as consensus analysis, multidimensional scaling and hierarchical clustering. Expressed rules for deciding among treatment alternatives, particularly whether to contact a physician, will be explored using ethnographic decision tree modeling techniques. Both approaches should yield information useful to planners and evaluators of health education programs aimed at increasing the match between need for services as defined by the medical profession, and actual use of medical services by this population.