Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) is a technique widely used for the quantification of mRNA transcription, It is well recognized that the reference genes used in RT-qPCR require appropriate validation to ensure that gene expression is unaffected by experimental conditions. The differentiation of bone mesenchymal stem cells (BMSCs) into neurons is important in the treatment of nerve injury. In gene expression analysis of the differentiation of BMSCs into neuronal cells by, the commonly used reference genes for RNA analysis are often selected without any preliminary evaluation of their suitability. The present study aimed to evaluate the mRNA expression levels of 11 putative reference genes, including ACTB, ARBP, B2M, CYCA, GAPDH, GUSB, HPRT, PPIA, RPL13A, TBP and PGK1, in order to select the most suitable reference genes for RT-qPCR of the differentiation of neuronal cells by BMSCs. The mRNA expression levels of the 11 putative reference genes were examined using RT-qPCR in rat BMSCs differentiated into neuronal cells. Normal BMSCs and three types of rat BMSCs, which were chemically induced to differentiate into neurons using neurotrophic cytokines and co-culture with retinal cells. The geNorm, NormFinder and BestKeeper software programs were used to select the most suitable reference genes. The results of the analyses using the three software programs demonstrated that RPL13A was the most stable among all the groups, while ACTB was the least stable. The combination of CYCA and PPIA reference genes contributed the most to increasing stability. The suitability of selected reference genes requires previous pre-selection in every investigation. Based on the three software programs, RPL13A, and the combination of CYCA and PPIA were identified as the most suitable reference genes for RT-qPCR in neuronal cells differentiated from BMSCs.