Developing mouse retina has been serving as an ideal model for investigating the molecular mechanism of neural development and angiogenesis, because several significant events associated with these physiological phenomena are drastically occurring in conjunction with retinal development. However, as many genes are influencing on each other to establish mature retina within 21 days from E10 to P12, we must carefully design the experiments, such as in the case of quantitating the amount of altered gene expression toward the establishment of retina by quantitative PCR. As we have seen considerable variations of quantitative results in different developmental stages of retina depending on the reference genes used for compensation, we here attempted to determine a reliable reference gene to accurately quantitate the target genes in each stage. According to the results of in silico prediction and comparison with a database of SAGE, we found that the most stable gene from early to late stages was Sdha, whereas one of the most popular housekeeping genes, Actb, was the one that could mislead the quantitative results even in the adult stage. Consequently, we pointed out the importance of selecting an appropriate reference gene, especially to quantitate the amount of gene expression in the developmental stages of a certain tissue. : The manuscript describes a significance of selecting appropriate reference genes for compensation in advance of performing qPCR to accurately quantitate the target gene expressions. As we have encountered considerable variations of quantitative results in different developmental stages of mouse retina, we carefully evaluated the reliability of reference genes by using in silico prediction program, as well as by comparing the quantitative results with a SAGE database. In fact, one of the eight examined genes, Sdha, was found to be reliable from early to late stages of retinal development. In contrast, Actb, which is known to be one of the most popular genes for qPCR compensation, was found to be unstable and could mislead the quantitative results throughout the stages.(Figure is included in full-text article.)