The use of reverse transcription (RT) PCR for relative quantitation of gene transcripts relies on the reproducibility of the individual RT, PCR and product measurement steps. Semi-competitive RT-PCR (RT-cPCR) uses an internal competitor template in the PCR step to improve quantitation. We have surveyed the reproducibility of RT, PCR, RT-cPCR and measurement, amplifying the glyceraldehyde-3-phosphate dehydrogenase "housekeeping" gene from isolated renal glomeruli. We used an enzyme-linked immunosorbent assay (ELISA) to quantify PCR products. We also report our PCR-based method for constructing a competitor DNA identifiable independently of the native product. Our results show that the entire RT-PCR and ELISA process had a standard deviation (SD) of less than 10% (n = 10). This compared to an SD of less than 13% (n = 10) in PCR and ELISA. The SD for ELISA alone was less than 11% (n = 10). RT-cPCR quantitation gave an SD of approximately 15% (n = 10). These results support the use of standard RT-PCR for the relative quantitation of mRNA. RT-cPCR is also suited to relative quantitation, but it is also independent of the amplification saturation curve and permits the identification of differences in cellularity between samples.