To better comprehend the relationship between perioperative blood transfusion and survival time after curative gastrectomy for advanced gastric cancer, the authors reviewed retrospectively data on 568 patients treated in their clinics from 1965 to 1983. Of these 568, 195 (34.3%) required no blood transfusion and 373 (65.7%) required transfusions within the perioperative period. Univariate analysis indicated that the survival time of the transfusion recipients was significantly less than that of the patients who had no transfusions (P less than 0.01). In subgroups of the authors' patients stratified to adjust for stage of disease, there was, however, no significant difference between the survival rates. Subsequently, multivariate analysis, using the Cox regression analysis, which adjusted for sex, age, and other covariates, indicated that perioperative blood transfusion was not a useful factor for predicting survival time. Multivariate analysis suggested that tumor size (P less than 0.01), degree of invasion into the gastric wall (P less than 0.01) and status of lymph node metastasis (P less than 0.01) were the most important covariates after curative gastrectomy for advanced gastric cancer. The authors' findings revealed the lack of any relationship between perioperative blood transfusion and survival time of patients who underwent curative resection for advanced gastric cancer.