Sentinel lymph node dissection is a minimally invasive surgical technique for staging of breast carcinoma. The optimal pathologic examination of the sentinel node (SN) has not yet been determined. Our standard protocol for evaluation of the SN in patients with breast cancer included frozen section at one level, plus paraffin sections at two levels, separated by 40 microm, and stained with hematoxylin and eosin and cytokeratin immunohistochemistry (IHC) at each paraffin section level. In the current study, we evaluated the use of step sections and cytokeratin IHC in 60 SNs (42 consecutive patients) that were tumor-negative on frozen section and hematoxylin and eosin staining at permanent section levels 1 and 2. The SN were reexamined with cytokeratin IHC at eight additional levels (levels 3-10) of the paraffin block, each separated by 40 microm. Previous IHC sections from levels 1 and 2 had shown micrometastases in nine SNs (eight patients) and no tumor cells in the remaining 51 SNs (34 patients). Of the 51 previously negative SNs, only two (4%) SNs from one (3%) patient had metastatic carcinoma cells in levels 3-10. Thus, the additional step sections with cytokeratin IHC did not significantly increase the number of patients with tumor-positive SNs. We currently recommend that the SN be examined with cytokeratin IHC at two levels of the paraffin block. This should optimize sentinel lymph node dissection as a staging technique and minimize the labor and financial burden associated with multiple step sections and IHC stains.