Considerable controversies exist regarding whether elderly patients with early-stage breast cancer receiving breast-conserving surgery (BCS) should forgo radiotherapy. We utilized the National Cancer Database to analyze data of 115 516 women aged ≥70 years, treated with BCS for T1-2N0-1M0 breast cancer between 2004 and 2014. Multivariable Cox proportional hazards model was used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for mortality 3, 5 and 10 years after 90 days of BCS associated with radiotherapy. Patients who received no radiotherapy had a higher mortality rate than those who received radiotherapy (5-year survival rate: 71.2% vs 83.8%), with multivariable-adjusted HRs of 1.65 (95% CI: 1.57-1.72) for 3-year mortality, 1.53 (1.47-1.58) for 5-year mortality and 1.43 (1.39-1.48) for 10-year mortality. The association held even for patients ≥90 years. This association was observed in all strata by reasons for radiotherapy omission, receipt of endocrine therapy or chemotherapy, calendar period and other clinical characteristics, with 40% to 65% increased 5-year mortality for patients without radiotherapy. This positive association persisted when analyses were restricted to patients with T1N0 and estrogen-receptor-positive disease who had received endocrine therapy (5-year mortality: HR 1.47 [1.39-1.57]) and in propensity score weighted analyses. Our study shows, in routine practice, elderly patients who received no post-BCS radiotherapy had higher total mortality than those who received radiotherapy. These findings suggest that the current recommendation of omission of post-BCS radiotherapy for elderly women with early-stage breast cancer may need to be reconsidered, particularly for those without contraindication.