Mammography use decreases with age although the risk of breast cancer increases with age. Medicare now provides biennial coverage for screening mammography. This study was designed to simulate the Medicare condition by subsidizing mammography among women in eight retirement communities in the metropolitan Philadelphia area. The study also measured the impact of health education interventions and the presence of a mobile mammography van on increased use of mammography. Retirement communities were assigned randomly to the control (cost subsidy alone) or experimental group (cost subsidy, mammography van, and tailored health education interventions). A total of 412 women ages 65 and older who had not had mammograms in the previous year were surveyed at baseline and 3 months later. Analytic techniques reflected the cluster nature of the randomization. Women in the experimental group were significantly more likely than the control group women to have obtained mammograms. Forty-five percent of the experimental group women compared with 12 percent of the control group women subsequently had mammograms in the 3 months after the baseline interview (P less than .001). Logistic regression analysis for mammography use indicated an odds ratio of 6.1 associated with being in the experimental group. For women in the experimental group, a separate logistic regression for mammography use showed an odds ratio of 7.8 associated with attendance at the educational presentation. The results suggest that Medicare coverage alone will not increase mammography use sufficiently to achieve year 2000 objectives. However, the addition of access enhancing and health education interventions boosts utilization dramatically.