PURPOSE: Perioperative blood transfusion and its influence on the immune system in cancer surgery is a subject of controversy. We made a retrospective study to comprehend the prognostic effects of perioperative blood transfusion in gastric cancer surgery. METHODS: A total 284 patients who underwent gastrectomy for gastric cancer from 1991 to 1998 were retrospectively reviewed. Uni- and multi-variated analyses of the incidence and amount of perioperative blood transfusion were performed, along with a comparison of the clinicopathologic features. RESULTS: Of the 284, 119 (42%) required no blood transfusion and 165 (58%) required blood transfusion within the perioperative period. The transfused group included patients with larger tumors (more than 4 cm, 67.1% vs 47.5%, P=0.001), with longer operation time (260.8 vs 229.2 min, P=0.001), with total gastrectomy (29.7% vs 14.4%, P=0.001), with advanced T-stages (P=0.001), and with more advanced nodal metastasis (P=0.005) than the nontransfused group. Overall comparison of transfused patients versus nontransfused patients by log rank analysis revealed a statistically significant adverse influence of blood transfusion on survival rate (58.7% vs 80.3%, P=0.001). However, after stratifying patients into stages and applying Cox-regression analyses, blood transfusion did not appear to have any effect on prognosis except stage III. CONCLUSION: We could not find any direct causal relationship between perioperative transfusion and long term prognosis in patients receiving gastric cancer surgery. However, in cases with advanced gastric cancer, it is better to refrain from unnecessary blood transfusion in the perioperative periods.