New surgical oncology techniques, such as lymphatic mapping and sentinel node biopsy, require precise identification of the presence of even very small numbers of tumor cells. The gold standard for such analysis remains microscopic assessment of tissue sections, stained conventionally or by immunohistochemistry for appropriate tumor markers. This approach is limited by sampling constraints and requires a high degree of expertise from the microscopist. Recent studies have demonstrated a subgroup of patients whose sentinel nodes are negative on microscopy, but whose nodes yield an enhanced signal for melanoma markers when evaluated by RT-PCR. These enhanced signals reflect a mixture of signal sources, including small numbers of melanoma cells and cells other than melanoma cells that express the relevant markers(s). Because the preparative techniques for RT-PCR destroy the structural integrity of the tissues and disrupt individual cells, the exact cellular source of enhanced signal from a tissue cannot be demonstrated by conventional RT-PCR. RT in situ PCR, in which the RT-PCR technique is applied on a tissue section, does identify the cells that are the source of signal. We have attempted to optimize this interesting approach and have applied it to the detection of relevant melanoma markers in tissue culture lines.