Increasingly, reverse transcriptase polymerase chain reaction (RT-PCR) is used to detect clinically significant tumour cells in blood or bone marrow. This may result in a redefinition of disease-free and clinical relapse. However, its clinical utility may be limited by lack of automation or reproducibility. Recent studies have suggested nucleic acid sequence-based amplification of target RNA may be more robust. In this study, nucleic acid sequence-based amplification was established to detect melanoma, colorectal and prostate cancer cells. Nucleic acid sequence-based amplification and RT-PCR both successfully amplified target RNA in peripheral blood samples from patients with melanoma and colorectal cancer, but only RT-PCR detected PSA in blood samples from patients with prostate cancer. There was relatively good agreement between sample replicates analyzed by RT-PCR (Kappa values of one for tyrosinase, 0.67 for CK-20 and one for PSA), but less agreement when analyzed by nucleic acid sequence-based amplification. This may limit the routine use of NASBA for the detection of clinically significant disease. In summary, RT-PCR appears at present to be the most reliable and reproducible method for the detection of low-level disease in cancer patients, although prospective studies are warranted to assess the clinical utility of different molecular diagnostic methods.