Pathologists are under increasing pressure to submit fresh tissue for ancillary studies and research protocols. In several tumor types (breast, lung, melanoma, colorectal, prostate), increased interest in detecting submicroscopic nodal metastases through reverse transcriptase polymerase chain reaction analysis of mRNA from portions of lymph nodes has precluded histologic analysis of the entire node for metastases. A retrospective review was undertaken of 227 breast cancer patients prospectively entered on a research protocol examining the usefulness of sentinel lymph node surgery. All of the patients ultimately underwent complete lymph node dissection. The research protocol required that all nodes greater than 8 mm in size be bisected and submitted separately. Positive lymph nodes were evaluated for unilateral or bilateral involvement in the node sections. Sixty node-positive patients were identified, yielding 230 positive nodes. One hundred seven of these nodes were confirmed to have been bisected. Carcinoma was identified in both lymph node sections in 64 (59.8%) nodes and in only one-half of the bisected lymph node in 43 (40.2%) nodes. Involvement of both sections was more likely when patients had multiple nodes positive. In 12 patients, involvement of one-half of the bisected nodes was the only evidence of metastatic disease (20.0% of node-positive patients). This evidence suggests that submission of less than the complete lymph node for histologic evaluation of metastatic disease decreases the accuracy of lymph node staging. Furthermore, a significant proportion of patients may be erroneously classified as histologically node negative.