Previous studies on the association between mammography screening and stage at breast cancer (BC) diagnosis have limitations because they did not analyze persistence with mammography screening and did not distinguish screening from diagnostic mammograms. The objective of this study is to determine the association between persistence with mammography screening and stage at BC diagnosis among elderly women. A retrospective observational study of 39,006 women age ≥70 diagnosed with incident BC from 2005 to 2009 from the Surveillance, Epidemiology, and End Results (SEER)-Medicare dataset was conducted. A validated algorithm with high sensitivity and specificity was used to distinguish between screening and diagnostic mammograms. Persistence with mammography screening was measured as having at least three screening mammograms in five years before BC diagnosis. Multinomial logistic regressions were performed to analyze the association between persistence with mammography screening and stage at diagnosis, in a multivariate framework. Overall, 46% of elderly women were persistent with mammography screening, 26% were not persistent, and 28% did not have any screening mammogram in five years before BC diagnosis. As compared to women who were not persistent with mammography screening, women who were persistent with mammography screening were significantly more likely to be diagnosed at earlier stages of BC. The adjusted odds ratios were 3.28, 2.37, and 1.60 for in situ, local, and regional stages, respectively. A lower proportion of elderly women was persistent with mammography and it was highly associated with earlier stages of BC diagnosis. Interventions designed to promote persistent mammography screening among elderly women are warranted.