Using reverse transcriptase-polymerase chain reaction (RT-PCR), we have recently described a bona fide deletion within the coding sequence of the large subunit of ribonucleotide reductase (R1) mRNA in colon cancer. Consecutive studies have raised questions about the nature of this phenomenon, because the corresponding genomic alteration at the DNA level or an aberrant protein could not be detected. Thus we considered an in vitro artifact during RT-PCR as a possible explanation for this observation. In contrast to reverse transcriptase, Taq DNA polymerase or C. therm DNA polymerase did not generate the aberrant product, suggesting the demand for the template switching activity intrinsic to retroviral reverse transcriptases. In fact, virtually the same deletion was observed in RT-PCR experiments when in vitro transcribed R1 mRNA was used. Considering structural prerequisites for template switching within R1 mRNA, we show that two direct repeats adjacent to a strong stem-loop secondary structure flank the deleted region of 1851 base pairs. Because several mRNAs encoding proteins of clinical and diagnostic importance fulfill these criteria, template switching enhances the potential risk of observing artifacts when interpreting results from RT-PCR studies. As shown in the present example, this may involve the artificial generation and the misinterpretation of PCR fragments amplified from targets relevant to tumor biology or cancer pharmacology. As a possible solution, one-step PCR with C. therm polymerase should be considered. This polymerase eliminates the artificial generation of aberrant mRNA signals observed during cDNA synthesis.