Background: Randomized controlled trial data support safe omission of axillary surgery in patients >70 years old with T1N0 hormone receptor-positive (HR+) breast cancer, but utilization of axillary surgery in the U.S. remains > 80%. We sought to assess the proportion of variance in axillary surgery receipt that is attributable to surgeon-level differences vs. patient level differences, and assess the patient- and surgeon-level factors that affect axillary surgery receipt in this population. Methods: Women >70 years old with T1N0 HR+/HER2-negative breast cancer diagnosed from 2013-2015 in SEER-Medicare were linked to surgeon-level data from the American Medical Association Masterfile. Patients of low-volume surgeons (surgeons treating < 5 patients in this population over the study period) were excluded. To assess patient-and surgeon-level factors associated with axillary surgery receipt, we estimated hierarchical logistic regression models regressing receipt of axillary surgery on surgeon-specific random intercepts (null model). We then re-estimated this model including patient factors (age, Charlson Comorbidity Index (CCI), and a validated claims-based frailty index). Our final model included other surgeon-, patient-, and disease-level factors. The intracluster correlation (ICC) was used to estimate the proportion of total variance in the outcome attributable to the surgeon level. Results: Of 4,410 women included, 58.3% were >75 years old, 20.7% were frail, and 13.9% had a CCI>2; 86.1% of women underwent axillary surgery. Of 432 surgeons represented, 52.6% were female and 9.5% identified as surgical oncologists. In the null model, 10.5% of the variance in the axillary evaluation was attributable to the surgeon-level, with 89.5% attributable to patient-level differences (Table). Adjusting for key patient-level variables (age, CCI, and frailty status) reduced unexplained patient-level variance, resulting in a greater proportion of total unexplained variance attributable to the surgeon level (13%). In our final model, physician subspecialty and years in practice were non-significant, but patients of female surgeons were less likely to undergo axillary surgery (OR 0.69, 95% CI [0.51-0.95]). Conclusions: In older women with T1N0 HR+ disease, patient-level differences reassuringly account for most of the variation in receipt of axillary surgery. Drivers of residual patient-level variation may include unmeasured factors such as differences in surgeon-patient communication and patient preferences.