Quantitative differences in the expression of oncogenes are a critical feature of the cancer process. Several methods are currently available for assessing differential gene expression, but none can be used to determine quantitative changes in gene expression from small numbers of cells. The ability to conduct this type of quantitative analysis would be useful in the study of definable, early stages of carcinogenesis when very few cells are involved. We therefore developed a highly sensitive, slide-based technique that incorporates the benefits of in situ polymerase chain reaction (PCR) and reverse transcription-PCR (RT-PCR) to quantify differential c-myc gene expression from liver tissue sections having either low or high levels of proliferating hepatocytes. To eliminate the need for isolating and quantifying mRNA, cells of interest were microdissected from frozen histological sections and their RNA directly subjected to RT-PCR amplification. These reactions were conducted in the presence of an internal RNA standard that was specifically designed to normalize differential RT and PCR efficiencies between samples. GENESCAN software analysis was used to determine the ratios of the RT-PCR products of the target gene to the RNA standard. These ratios were then normalized to the numbers of cells isolated, as quantified by image analysis, and comparative gene expression values were determined between sample groups. We conclude that this technology can be adapted to study any gene of interest in any type of frozen tissue or isolated cells. This methodology is particularly applicable to the molecular analysis of histopathologically distinct preneoplastic and neoplastic lesions identified on tissue sections.