Normally, qRT-PCR is based on relative quantifications that consist of normalizing the target gene with an internal control, which is a gene that presumably maintains stable expression during the experiment and is termed a housekeeping gene. HKGs have been validated for several experimental models and tissues; however, proper precautions are not always taken to account for their variabilities, thereby compromising the reliability of the data. Although their uses have been well-established, they have been demonstrated to possess low stability in an in vitro hypoxia model. Our group has previously validated the optimal HKGs for use with other sleep impairment models, but the effects of sleeprelated breathing disorders, such as OSA, on HKGs have not yet been studied. Considering the increasing number of studies involving CIH models, including those using genetic approaches, and particularly qRT-PCR, we aimed to validate HKGs for use in studies involving the commonly used CIH model. qRT-PCR is the most commonly used method for the quantification of mRNA; however, its reliability depends on the correct normalization of the results using stable HKGs, a bad choice of which can compromise the results. There have been several studies involving HKGs that have been performed using different hypoxia models, but most of them have been conducted in vitro with an acute hypoxia model. These studies are usually conducted by subjecting a specific cell line to differing oxygen concentrations. In some of them, commonly used HKGs were observed to exhibit altered expression levels in hypoxic conditions. The current study showed that all of the analysis software programs produced similar results, particularly geNorm and NormFinder. For the analysis using BestKeeper software, the only difference was observed in the temporal cortex, in which different optimal candidates were revealed compared with those identified by the other 2 software programs. NormFinder and geNorm presented similar HKG ranks, the similarity may be due to pairwise comparison methods used in both, for the same reason, BestKeeper presents different ranks, when compared to NormFinder and geNorm, due to comparison method by Pearson correlation ) that is to rank HKG. Our data indicate that all of the candidate genes that were tested are suitable for use according to the adopted cut-off values. The only exception is 18S, which was the least stable gene in almost all of the structures, independent of the method of evaluation, in contrast with previous studies involving in vitro hypoxia. Thus, 18S is not advisable as an HKG under any conditions, due to poor ranking position in most structures. Our data demonstrated that 18S was not stable following CIH exposure, corroborating previous studies reporting that its expression varies according to the cell line that is being tested under hypoxic conditions. 18S has been shown to be stable in HEK and PNT2 cells, but to also be the least stable in LNCap and MCF-7 cells. Interestingly, 18S stability varies in the PNT2 and LNCap cells in a contrasting manner; both are present in the same tissue type but under different conditions; i.e., physiological and pathological conditions, respectively. A study of the brain of an in vivo model of CIH revealed that 18S stability is homogeneous, demonstrating the sensitivity of a majority of brain cells to the CIH model in all structures. No data are currently Perifosine available in the literature describing HKGs in specific brain structures using hypoxia models, and most of the studies have been performed in vitro.