Lymphatic mapping and sentinel lymph node (SLN) biopsy are widely used as a staging technique for patients with cutaneous malignant melanoma who are at risk for metastases. SLN status has been shown to be a strong predictor of prognosis, and a variety of techniques have been used to identify minimal metastatic disease in SLNs. However, there is no validated consensus method for the optimal histologic analysis of SLNs harvested from melanoma patients. This study was conducted: 1) to assess the yield of metastatic melanoma detected in SLNs deemed negative by initial routine pathologic analysis (RPA) by subjecting them (after review of the original slides) to enhanced pathologic analysis (EPA) that included complete step-sectioning and immunohistochemistry (IHC); 2) to characterize the distribution of metastatic melanoma deposits within the SLNs; 3) to determine a preferred method of pathologic analysis applicable to daily practice; and 4) to attempt to assess the clinical significance of disease detected by EPA. A total of 105 SLNs were harvested from 49 patients who underwent successful SLN biopsy procedures during the period of study. Ten SLNs from 10 patients were positive on initial RPA and were not analyzed further. Ninety-five SLNs from the remaining 39 patients were reviewed and processed with additional hematoxylin and eosin, S-100 protein, and HMB-45 stains at 50-microm intervals for 20 levels or until the SLN tissue was exhausted. A single pathologist reviewed all sections without knowledge of the results of the other stains. Overall, metastatic melanoma was discovered in SLNs from 20 of the 39 patients: SLNs from 6 patients were found to have melanoma on review of the original hematoxylin and eosin slides, and SLNs from 14 patients were positive only after EPA. Twenty-one individual positive SLNs from these 14 patients were detected by EPA; of these, 10 positive SLNs were identified solely by IHC, representing 12% of the patient cohort and 10% of all SLNs studied by EPA. Detection rates were significantly associated with the staining method and the number of levels performed (P < 0.01). S-100 protein staining resulted in the highest yield of SLN positivity (86%), followed by HMB-45 (81%) and hematoxylin and eosin (52%). No single method detected all of the micrometastases. A detailed topographic mapping of metastatic deposits in SLNs was carried out. When using all three staining techniques, all 20 levels were required to identify 100% of the micrometastases; 95% of positive SLNs were identified with 17 levels, 90% with 15 levels, 75% with 10 levels, and 42% with 3 levels. Projected rates of detection for various different sectioning strategies were determined, with alteration of either the number of levels examined, the interval between the levels, or both. Detection of SLN positivity can be increased to 71% by performing three levels at 250-mum intervals, each level being composed of a set of three sections stained with hematoxylin and eosin, S-100 protein, and HMB-45, respectively. Therefore, this is the methodology we propose for the study of SLNs in melanoma patients. After a median follow-up of 87 months (range, 9-134 months), patients with EPA-detected disease and those with negative SLNs by EPA demonstrated improved recurrence-free and disease-specific survival compared with patients with RPA-detected disease in SLNs. Sampling error introduced by variations in pathologic processing should be addressed by standardization of pathologic methods, and the clinical significance of minimal SLN disease should be addressed in prospective studies of homogeneously staged patients.