There is no consensus method for the histological analysis of axillary sentinel nodes (SN). This study aimed to (1) assess the rate of occult metastases in SN using large serial sectioning and immunohistochemistry (IHC), (2) evaluate whether occult metastases were predictive of metastases in the downstream axillary nodes, and (3) specify a methodology of analysis of SN that could be both sensitive and applicable in daily practice. One hundred three patients with breast carcinoma underwent SN biopsy and then axillary dissection. SN free of tumor at standard examination of one section were sectioned at six levels (150-microm intervals) and immunostained for cytokeratin. The number and localization of labeled metastatic cells (occult metastases) were recorded. In 29 of the 103 patients (28%), SN were found to be metastatic after standard examination. The SN of the remaining 74 patients were further analyzed using IHC. Occult metastases were detected in 35 of these patients (47.3%), leading to an overall SN involvement rate of 62% (29+35/103). In 33 of these 35 cases, the plurality and the dispersion of the immunostained cells implied that the screening of only 3 of the 6 levels would have led to the detection of diagnostic positive events. Only one of the 35 patients (2.8%) with occult metastases showed metastatic lymph node in the downstream axilla. In our series of axillary SN, the analysis of one standard histologic section and, when negative, of only three additional sections after IHC revealed >60% of metastasis or occult metastasis. Metastasis detected by standard analysis had a high predictive value of downstream node metastasis, whereas the predictive value of occult metastasis revealed by IHC was poor. The clinical significance of occult metastases in SN needs to be specified by long-term follow-up analysis.