Quantitative analysis of the leukemia fusion gene by real-time PCR is a sensitive method to monitor minimal residual disease; the data obtained are very useful to evaluate the disease stage and prognosis, contributing to the clinical practice of hematology. However, there is no standard laboratory procedure for leukemia genetic testing. Therefore, this genetic testing has some problems related to precision management. To minimize analytical error factors, normalization by an internal control gene is necessary. Additionally, it is important to choose a gene suitable for leukemia gene expression analysis because the expression of an internal standard gene changes due to various factors. In this study, we examined analytical error factors (RNA extraction efficiency, reverse transcription reaction efficiency) and evaluated an internal control gene. As a result, in RNA extraction, the extraction efficiency of the acid-guanidium-phenol-chloroform (AGPC) method was high compared to the silica method. The reverse transcription reaction efficiency was significantly different with each reaction reagent. Furthermore, since three kinds of gene (18s rRNA, GUS, beta-actin) had few differences between samples, they were considered to be suitable as internal standards.