Conflicting results have been obtained by various research groups using tyrosinase reverse transcription-polymerase chain reaction (RT-PCR) for detecting micrometastases in the blood of melanoma patients, with positive results ranging from 0 to 100% in disseminated melanoma. Methodological differences in the processing of blood samples may in part account for these discrepancies. The aim of this study was to standardize and optimize the experimental conditions for RT-PCR detection of melanoma cells in peripheral blood. We analysed the effect of different parameters of sample processing on the sensitivity of the tyrosinase RT-PCR using peripheral blood spiked with defined numbers of SKMEL28 melanoma cells. Purification of the mononuclear cell fraction using a Ficoll gradient with a density of 1.077 g/mL prior to RNA isolation gave the highest sensitivity, with the detection of two SKMEL28 cells in 5 mL of blood. In addition, the RNA isolation method and the kind of RT enzyme used had a significant impact on the sensitivity and reproducibility of tyrosinase detection, whereas variations in the PCR conditions had only a minor influence. Furthermore, we showed that amplification of MelanA in addition to tyrosinase resulted in an approximately 10% enhanced sensitivity of melanoma cell detection, whereas gp100/pMel17 and MUC18 gene products were also detected in blood from non-melanoma patients. MelanA could serve as a sensitive marker in addition to tyrosinase for detecting micrometastases.