This study examined the impact of individual demographic characteristics (age, race/ethnicity, and type of reporting hospital), together with measures of social context, including area of residence socioeconomic status (SES), change in SES, and access to screening (area mammography capacity), on breast cancer stage at diagnosis among New York City residents diagnosed between 1980 and 1985. Three logistic regression models evaluated the impact of individual variables, social context variables, and both classes of variables combined on the outcome of having late-stage (regional or distant) compared to early-stage (local) cancer. The logistic regression models indicated that women aged 50 to 64 years were independently more likely to have late-stage cancer than those younger or older. Controlling for individual and social context variables, African American women were 25 percent more likely than White women to have late-stage breast cancer (p < 0.0001); higher odds seen in the individual model for Hispanic women disappeared after controlling for area SES. In the combined model, women diagnosed from public hospitals were 44 percent more likely to have late-stage disease than those diagnosed in nonpublic settings (p < 0.0001). In both the social context and combined models, area mammography capacity was a significant independent predictor of stage (p = 0.016); area-level SES, but not change in SES, was independently related to late-stage disease (p = 0.002). These data suggest that breast cancer control activities should more actively ensure adequate access to screening, particularly for minorities, populations living in socioeconomically disadvantaged areas, and those cared for in the public sector.