The invention belongs to the field of genetic testing and provides a peritoneal metastasis prediction model of a gastric cancer based on 22 genes and application thereof, including PCLO, UGGT1, ZNF714, KIAA0825, COL23A1, MED1, NPAS2, TTC14, RPS27A, ASPH, ARHGEF12, SIK1, PAPPA, HHIPL1, MYO9B, ITPKB, ZNF862, MKNK1, MUC6, TRRAP, DUOX1 and KRTAP5-2; a risk of peritoneal metastasis is predicted effectively and specifically, according to SVM of a selected classifier and a positive judgment threshold value of 0.5. The application of the gene model of the invention is helpful for predicting the metastasis of patients with gastric cancer and has important value and significance for taking timely and effective clinical measures, making an individualized diagnosis and treatment plan and finally improving a survival rate of the patients with the gastric cancer.