Quantitative real time reverse transcription polymerase chain reaction has been applied in a vast range of studies of gene expression analysis. However, real-time PCR data must be normalized with one or more reference genes. In this study, eleven putative consistently expressed genes (ACT, TUA, TUB, CYP, DNAj, ELFA, F-box27, RPL12, GAPDH, UBC and UBQ) in nine Siberian Apricot Germplasms (including much variability) were evaluated for their potential as references for the normalization of gene expression by NormFinder and geNorm programs. From our studies, ACT, UBC, CYP, UBQ and RPL12 as suitable for normalization were identified by geNorm, while UBC and CYP as the best pair by NormFinder. Moreover, UBC was selected as the most stably expressed gene by both algorithms in different Siberian Apricot seed samples. We also detected that a set of three genes (ACT, CYP and UBC) by geNorm as control for normalization could lead to accurate results. Furthermore, the expression levels of oleosin gene were analyzed to validate the suitability of the selected reference genes. These obtained experimental results could make an important contribution to normalize real-time PCR data for gene expression analysis in Siberian Apricot Germplasm.