According to standard-protocols, real time quantitative RT-PCR (RQ-PCR) for quantification of BCR-ABL fusion transcripts in CML patients is performed with the construction of a standard curve for each run, each sample is analyzed at least in duplicate and 10-40 ml peripheral blood are processed. This approach is appropriate for a research laboratory, but is not suitable for a routine laboratory setting. We show that the calibration curve based on the common 5 dilution standards (between 10 and 10(6) copies) is strongly influenced by the large variability of the measurements below 100 copies of the gene. In other words, including a standard with 10 copies is a source of error, which cannot be reduced through the construction of a standard curve with each run. Adding additional dilutions between 10 and 100 copies to the standard curve, the variance of the obtained curve is much reduced. As a conclusion, it is unnecessary to construct a calibration curve with each run since only negligible inaccuracy of calibration is added to the inaccuracy of measurement. Running of the samples in duplicate seems unnecessary since the inaccuracy of the method can be correctly estimated. Finally, we propose a standardized collection and isolation of total RNA from only 2.5 ml blood using an integrated system, which allows RNA stabilization for up to 5 days and provides snapshots of BCR-ABL fusion transcripts with higher accuracy than with non-stabilized blood samples.