The polymerase chain reaction (PCR) theoretically offers a very powerful means of quantifying gene expression in cells and tissues of different histological classes. From a technical viewpoint however, the use of RT-PCR to quantify gene expression can be demanding with poor reproducibility arising from several diverse sources. In this study, we describe and fully characterise an RT-PCR assay which generates highly reproducible estimates of DNA polymerase beta gene expression relative to the expression of beta-actin in a semi-quantitative manner. Particular emphasis has been placed on the efficiency of first strand cDNA synthesis and to aspects of primer design. In addition, various aspects associated with the quantification of gene expression required to generate reproducible results are discussed. Using the techniques described herein, the quantification of DNA polymerase beta expression in three independent experiments using human colon carcinoma cells was highly reproducible with ratios of target to internal standard gene expression of 3.68x10(-4) +/- 0.23x10(-4). Provided that careful consideration is given to key areas of the RT-PCR assay during experimental work up procedures, this assay can be used to provide an accurate measure of gene expression in cell lines.