Older women of color experience an elevated breast cancer mortality risk (2.2 times greater for blacks than whites) that is primarily the result of racial disparities in stage of diagnosis. Adherence to mammography guidelines is lowest for those women at highest risk. Improved utilization could prevent thousands of breast cancer deaths each year, but little is known about the specific effects among older women of cultural factors and barriers related to minority status. In particular, interrelationships among sociodemographic variables and structural factors responsible for non-compliance have not been adequately investigated. This study was designed to assess the relative contributions of three separate domains that affect mammography utilization: individual decision-making and resources, provider referral behavior, and institutional access. A cross-section survey of mammography behaviors has been developed with assistance from Women of Color Coalition for Health and piloted in the Emergency Department (ED) at Boston City Hospital, a public inner-city hospital affiliated with Boston University School of Medicine. Face-to-face interviews follow an eleven question format that investigates health beliefs, demographic factors, level of acculturation, regular source of care, access to other types of preventive health care and mammography utilization patterns. Group differences related to age and race/ethnicity will be assessed in a large, multi-ethnic sample of women greater than 50 years of age presenting to the ED for medical care. Multinomial Logit will be used to compare 3 groups: 1) no mammography ever; 2) past mammography; and 3) recent mammography. Nested Logit will be utilized to demonstrate the relative importance of different variables for two related but not identical decisions: baseline mammography and regular compliance. Medicare Part B claims-level data for mammography utilization, health status and provider type will also be investigated among women residing in the same zip code areas as survey respondents, using similar analysis techniques, and accuracy of self-report of mammography estimated. Results will help establish policy direction. If health beliefs and information are the principal determinants of mammography utilization, both public and private efforts can be targeted toward developing innovative, culturally competent educational outreach programs. If professional referrals are key to mammography compliance, then funds should be directed toward professional training, protocols and guidelines, and quality monitoring. If the primary problem is access to health care, changes will be required at institutional and state policymaking levels. Clear priorities must be set in order to meet the Healthy People 2000 goal of reducing needless breast cancer death.