Reverse transcription-polymerase chain reaction (RT-PCR)-based assays detecting occult neoplastic cells are increasingly being used for the study of tumour dissemination and minimal residual disease. However, different methods are employed by various research groups and the results are heterogenous. We prospectively assessed the results from nine laboratories performing tyrosinase RT-PCR assays for the detection of melanoma cells on a series of blind samples. After complete analysis, the results were compared for sensitivity and specificity. All laboratories reported correct results for cDNA standards. Five laboratories attained acceptable specificity and a sensitivity detecting 10 cells in 10 ml of whole blood. Four laboratories had unacceptable specificity and/or sensitivity. This blind study highlights the difficulty of RT-PCR data interpretation and the need for quality assurance between laboratories. Measures to increase the reliability of RT-PCR assays are proposed, which have to be prospectively evaluated in future studies.