For an exact comparison of mRNA transcription in different samples or tissues with real time quantitative reverse transcription-polymerase chain reaction (qRT-PCR), it is crucial to select a suitable internal reference gene. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and beta-actin (ACTB) have been frequently considered as house-keeping genes to normalize for changes in specific gene expression. However, it has been reported that these genes are unsuitable references in some cases, because their transcription is significantly variable under particular experimental conditions and among tissues. The present study was aimed to investigate which reference genes are most suitable for the study of gastric cancer tissues using qRT-PCR. 50 pairs of gastric cancer and corresponding peritumoral tissues were obtained from patients with gastric cancer. Absolute qRT-PCR was employed to detect the expression of GAPDH, ACTB, RPII and 18sRNA in the gastric cancer samples. Comparing gastric cancer with corresponding peritumoral tissues, GAPDH, ACTB and RPII were obviously up-regulated 6.49, 5.0 and 3.68 fold, respectively. Yet 18sRNA had no obvious expression change in gastric cancer tissues and the corresponding peritumoral tissues. The expression of GAPDH, β-actin, RPII and 18sRNA showed no obvious changes in normal gastric epithelial cells compared with gastric cancer cell lines. The carcinoembryonic antigen (CEA), a widely used clinical tumor marker, was used as a validation gene. Only when 18sRNA was used as the normalizing gene was CEA obviously elevated in gastric cancer tissues compared with peritumoral tissues. Our data show that 18sRNA is stably expressed in gastric cancer samples and corresponding peritumoral tissues. These observations confirm that there is no universal reference gene and underline the importance of specific optimization of potential reference genes for any experimental condition.