Background. Pathologic evaluation may lack the sensitivity required for accurate staging of the axilla in breast cancer patients. We have completed enrollment of a multi-institutional prospective cohort study designed to determine if molecular analyses can improve axillary staging. In subset analyses, we have attempted to address the following questions: (1) Does molecular analysis improve the sensitivity of sentinel lymph node biopsy (SLNB) and (2) is the sentinel lymph node (SLN) hypothesis valid at the molecular level? Methods. Four hundred eighty-nine subjects with T1, T2, or T3 breast cancer and no evidence of axillary lymph node (ALN) involvement were enrolled. ALNs were analyzed by routine pathology (hematoxylin-eosin staining), and by multimarker real-time reverse transcriptase-polymerase chain reaction analysis CRT-PCR to detect breast cancer metastases. Pathology and molecular data for both SLNs and nonsentinel ALNs were available for a subset of 207 subjects. Results. The sensitivity of pathologic analysis of the SLN to predict the pathologic status of ALNs was 84.1%. The sensitivity of combining pathologic with molecular analysis of the SLN to predict the pathologic status of ALNs was 92.8%, a statistically significant increase in sensitivity (P = .031 by the McNemar test for correlated proportions). Finally, the sensitivity of combining pathologic with molecular analysis of the SLN to predict the pathologic or molecular status of ALNs was 85.4%. Conclusions. The combination of pathologic and molecular analysis of SLNs resulted in the highest sensitivity for prediction of the pathologic status of ALNs. The data also provide evidence that the SLN hypothesis remains valid at the molecular level.