Objective: To explore the method of increasing the proportion of epithelial cells in the single-cell suspension of tissue in single-cell transcriptome sequencing in gastrointestinal epithelial tumors for improving the quality of test samples. Methods: The proportion of epithelial cell in the single-cell transcriptome sequencing data set of common gastrointestinal epithelial tumors (gastric cancer, colorectal cancer and pancreatic cancer) in GEO public database was analyzed. The difference in proportion and activity of epithelial cells in single-cell suspension between traditional enzyme digestion method and mechanical physical blowing method (improved method). Results: The data analysis in public database showed that the proportion of epithelial cells in 4 samples of normal gastric epithelium or gastric cancer was 17.05%, 6.11%, 8.93% and 14.66% respectively. The proportion of epithelial cells in 2 samples of colorectal cancer was 3.86% and 16.60%. The proportion of epithelial cells in 2 samples of pancreatic cancer was 3.22% and 21.37%. For two freshly collected gastric cancer samples, the cell agglomeration rate of single-cell suspension prepared by traditional enzyme digestion method was 56.26%±1.98% and 38.34%±1.26%, and the percentage of living cells in cell activity test was 98.43%±0.56% and 97.24%±0.48%, respectively. However, in the single-cell suspension treated by traditional enzyme digestion and mechanically blown with 0.33 mm insulin syringe for 30 seconds before being put on the machine, the epithelial cell agglomeration rates in 2 test samples were reduced to 22.78%±1.38% and 14.46±0.92%(P<0.000 1), and the proportion of living cells was 95.16%±0.42% and 93.52%±0.82% respectively(P<0.05). Conclusions: The pretreatment of tissue samples with traditional enzyme digestion and mechanical physical blowing could reduce epithelial cell agglomeration and increase the proportion of epithelial cells in single-cell suspension. The proportion of living cells decreased slightly, while it could meet the requirements of subsequent sequencing.