BACKGROUND: Sentinel lymph node (SLN) biopsy has been well-established for axillary lymph node staging for patients with breast cancer. For lymph node-negative patients, planned "backup" axillary lymph node dissection (ALND) is rarely indicated. Among patients with negative SLNs, the authors observed variation by tumor size and patient age in the total number of lymph nodes removed (SLNs plus non-SLNs). They hypothesized that this variation is an unrecognized source of bias for studies examining the morbidity of SLN biopsy. METHODS: Retrospective review of this institution's SLN database identified 4103 SLN biopsy procedures between 1997 and 2004 in which SLN biopsy was performed for prophylactic mastectomy, ductal carcinoma in situ, or T1 to T2 invasive cancers, and the SLNs were benign. RESULTS: The mean number of SLNs, non-SLNs, and total lymph nodes for all tumor sizes was 2.8, 1.5, and 4.3, respectively, and increased with tumor size (more lymph nodes were removed for T2 than for T1 tumors: 6.3 vs 4.3; P < .0001). This trend remained significant even in the later years of these investigators' experience with SLN biopsy, and was observed for 5 of 9 (56%) surgeons. More total lymph nodes were also removed in patients aged <or=50 years than in those aged >50 years (4.6 lymph nodes vs 4.2 lymph nodes; P = .006). In approximately 8% of patients (322 of 4103 patients), >or=10 lymph nodes were removed. CONCLUSIONS: The morbidity of SLN biopsy is less than that of ALND, but for pN0 patients, the total number of lymph nodes removed increased with tumor size and younger patient age. This variation is a previously unrecognized source of bias for studies that examine the morbidity of SLN biopsy.