Quantitative polymerase chain reaction (qPCR) analysis is a commonly used method for the study of mRNA expression throughout the field of mesenchymal stem cell (MSC) research. This technology is simple and sensitive; however the results may vary significantly due to the use of various reference genes (RGs) as normalizers. Therefore, the reliable use of RGs is vital for obtaining accurate results. The present study focuses on ten putative RGs for the normalization of qPCR data between human bone marrow-derived MSCs (BM-MSCs) and fetal tissue-derived MSCs (FT-MSCs). The total RNA from these two types of MSC was isolated using TRIzol reagent. cDNA was generated from the RNA via reverse transcription and subsequently analyzed by qPCR using ten common RGs as normalizers. These RGs included 18S, ACTB, B2M, HPRT1, GAPDH, TBP, PPIA, RPLP0, PGK1 and RPL13A. GeNorm, Norm Finder and Best Keeper software were used to analyze the qPCR results by evaluating the expression stabilities of the ten candidate RGs in BM-MSCs and FT-MSCs. Consequently, several of the commonly used RGs, including 18S, ACTB and TBP, were demonstrated to be unsuitable for normalization in these two MSCs, whereas RPL13A, B2M and PPIA were the most stable RGs and were therefore reliable for use in qPCR studies. Combining multiple RGs had no contribution towards increasing their stabilities. In conclusion, the present study revealed that RPL13A,B2M and PPIA were the optimal RGs for qPCR studies comparing BM-MSCs and FT-MSCs.