The ability of commensal bacteria to influence gene expression in host cells under the influence of pathogenic bacteria has previously been demonstrated, however the extent of this interaction is important for understanding how bacteria can be used as probiotics. Real-time quantitative polymerase chain reaction is the most sensitive tool for evaluating relative changes to gene expression levels. However as a result of its sensitivity an appropriate method of normalisation should be used to account for any variation incurred in preparatory experimental procedures. These variations may result from differences in the amount of starting material, quality of extracted RNA, or in the efficiency of the reverse transcriptase or polymerase enzymes. Selection of an endogenous control gene is the preferred method of normalisation, and ideally a proper validation of the gene's appropriateness for the study in question should be performed. In this study we used quantitative polymerase chain reaction data and applied four different algorithms (geNorm, BestKeeper, NormFinder, and comparative ΔCq) to evaluate eleven different genes as to their suitability as endogenous controls for use in studies involving colonic (HT-29) and vaginal (VK2/E6E7) human mucosal epithelial cells treated with probiotic and pathogenic bacteria. We found phosphoglycerate kinase 1 to be most appropriate for HT-29 cells, and ribosomal protein large P0 to be the best choice for VK2/E6E7 cells. We also showed that use of less stable reference genes can lead to less accurate quantification of expression levels of gene of interest (GOI) and also can result in decreased statistical significance for GOI expression levels when compared to control. Additionally, we found the cell type being analysed had greater influence on reference gene selection than the treatment performed. This study provides recommendations for stable endogenous control genes for use in further studies involving colonic and vaginal cell lines after bacterial challenge.