The yak is primarily found throughout the Tibetan high plateau and the surrounding mountainous area of south central Asia; among its others attributes, its milk is very important for the local population. A key concern in the field of yak research is the better understanding of which genes control the production and composition of milk. The most accurate and sensitive method for gene expression analysis is quantitative reverse transcription polymerase chain reaction (RT-qPCR). It is essential for reliable RT-qPCR to be able to the normalize the data using internal control genes (ICGs). However, it is critical to assess the reliability of the normalization by testing multiple ICGs. Our objective was to uncover a reliable normalization for RT-qPCR data obtained from yak mammary tissue during the lactation cycle. We assessed the reliability of 10 ICGs (ACTB, EIF6, GAPDH, LRP10, MRPL39, MRPS15, MTG1, RPS8, RPS23, and UXT) using geNorm. The analysis revealed that all of the tested ICGs can be considered to be reliable, but the use of the 6 most stable ICGs should be applied to yield a reliable normalization factor (NF). We compared the results of 3 target genes (CSN1S1, ESR1, and MYC) normalized using 6, 3, or 1 of the best ICGs. We did not observe overall differences between the 3 normalization strategies with the exception of 1 time point in MYC. The use of only a single ICG is not recommended; thus, we concluded that the calculation of the NF using the 3 best ICGs, MRPS15, RPS23, and UXT, is a reliable normalization strategy for RT-qPCR data obtained from yak mammary tissue during pregnancy and lactation. A dilution effect of the ICGs due to a large increase in the mRNA of abundantly expressed genes in bovine and porcine mammary tissue during the lactation cycle was previously observed. To test for the presence of a dilution effect in our study, we evaluated the pattern of non-normalized RT-qPCR data of ICGs from pregnancy to lactation and compared them with the total RNA concentration, milk yield, and non-normalized RT-qPCR data of 3 target genes. With a few exceptions, the non- normalized RT-qPCR data for the tested ICGs was significantly increased by lactation and had a positive correlation with total RNA and the non-normalized RT-qPCR data of CSN1S1. These data clearly indicated the presence of a "concentration effect" of single mRNA that remains unexplained but needs to be accounted for during the normalization of RT-qPCR data. Based on our findings, we recommend that the NF of the MRPS15, RPS23, and UXT genes should be used in the normalization of RT-qPCR data obtained from mammary tissue of lactating yaks during pregnancy and lactation.