Background: Volume of disease in the sentinel lymph node (SLN) is a significant predictor of additional nodal metastasis. This study assesses incidence of residual non-SLN disease in a large cohort of women with minimal SLN metastases and compares three methods of SLN micrometastasis volume measurement to determine which best predicts residual disease on completion axillary lymph node dissection (cALND). Methods: A total of 505 patients with invasive breast cancer and minimal SLN metastasis (pN1mi or pN0(i+)) underwent cALND and had complete data. All SLNs were evaluated by three measurement methods for volume of metastasis: (1) method of detection (frozen section, routine hematoxylin and eosin, serial hematoxylin and eosin, immunohistochemistry), (2) American Joint Committee on Cancer's AJCC Cancer Staging Manual, 7th edition, N category, and (3) number of metastatic cells (1-100, 101-999, ≥1000). Multivariable logistic regression models were used to predict the presence of additional non-SLN disease. Results: A total of 251 patients (50%) had pN0(i+) and 254 patients (50%) had pN1mi disease. Twelve percent of those with pN0(i+) and 20% with pN1mi had additional non-SLN disease. On multivariate analyses including eight variables, only lymphovascular invasion (odds ratio >2.2, P < 0.01) and volume of nodal metastasis as assessed by any method of measurement (method of detection, AJCC, and cell count) were significantly correlated with additional non-SLN disease (P = 0.04, 0.03, and 0.02, respectively). All three models had similar goodness of fit and discrimination (Akaike information criterion = 442, 442, 441; -2log likelihood = 416, 420, 417; concordance index = 0.680, 0.675, 0.676, respectively). Conclusions: A significant proportion of women with minimal SLN metastases have additional non-SLN disease at cALND. Assessments of SLN volume of disease by three different methods of measurement are equivalent for prediction of additional non-SLN metastases. © 2010 Society of Surgical Oncology.