The present studies demonstrate a theoretical and practical framework for the accurate quantitation of gene expression in RNA extracted from microscopic tissue samples. The approaches are developed around competitive RT-PCR techniques. Assay performance has been examined and validated at both the RT and PCR steps. Our analysis of RT transcription efficiency for a number of native and competitor combinations shows that this property can differ, even for very similar templates. However, this difference is consistent and, once identified and measured, can be removed as an obstacle to accuracy. Using mathematical modeling, we have examined the simulated co-amplification of native and competitor templates in PCR. Useful insights have emerged from such modeling which indicate that differences in initial amplification efficiency and the rate of decay of amplification efficiency during the reaction can rapidly lead to inaccuracy, even while the slope and linearity of log plots of the competitor input and reaction product ratios are close to ideal. Finally, we show here that competitive RT-PCR reactions do not have to remain in the log-linear phase of PCR in order to accomplish accurate and precise quantification. Using appropriate competitors sharing primer binding sites and high internal sequence similarity, identical amplification efficiencies are preserved throughout the reaction. Reaction products, including heteroduplexes formed between native and competitor templates as reactions progress to plateau, can be identified and quantified accurately using the new technique of denaturing HPLC (dHPLC). This analytical technique allows the accuracy of competitive RT-PCR to be preserved beyond the linear phase. The technique has high sensitivity and precision and target abundances as low as 100 copies could be reliably estimated.