Abstract: Objective:To explore whether vasculogenic mimicry (VM) exists in gastric adenocarcinoma (GAC) and to in -vestigate the clinicopathologic significance of VM in GAC. Methods:We tended to illuminate the mechanism of VM by per -forming immunohistochemical staining of MMP- 2, MMP- 9 and Cathepsin D. A total of 173 GAC samples with detailed fol-low-up data were collected. CD31/ periodic acid-Schiff (PAS) double staining and CK 8 ＆18immunohistochemical staining were performed to validate the existence of VM in GAC. The values of MVD (microvascular density) and VMD (vasculogen-ic mimicry density) were counted respectively. Immunohistochemical staining of MMP- 2, MMP-9 and Cathepsin D was per-formed for all samples. Results: VM was observed in40of the 173 GAC samples, especially in poorly differentiated GAC (P=0.014 ). Patients with VM were prone to hematogenous metastasis and distant recurrence compared with those without VM (P=0.020 , 0.029 ). Higher VMD count was also associated with hematogenous metastasis ( P=0.003 ). There was not significant difference in MVD count between VM-positive and VM-negative groups ( F=1.596 , P=0.482 ). The Kaplan-Meier sur -vival analysis showed that the survival duration of the VM-positive group was significantly shorter than that of VM-negative group (P=0.022 ). Cox proportional hazards model indicated that the VM and TNM stage were independent predictors for poor prognosis of GAC (P=0.039 and 0.004 ). The immunohistochemical expression of MMP-2, MMP- 9 and Cathepsin D was higher in VM-positive group than in VM-negative group (P<0.05). Conclusion:VM exists in GAC, especially in poorly differentiated GAC. VM is an unfavorable prognostic indicator for GAC. MMP-2, MMP- 9 and Cathepsin D might involve the formation of VM in GAC.