The use of perioperative blood transfusion (PBT), the immunological status pre-operatively and at discharge from hospital, and the clinical course were examined retrospectively in 124 patients who underwent 'curative' resection for gastric cancer at Shinkokura Hospital, Japan from 1979 to 1988. The general condition of patients with PBT was worse than that of those without PBT and the pre-operative immunological status of patients with PBT was less favourable than that of those without PBT. At the time of discharge from hospital the immunological condition remained worse for patients who had been given PBT. The clinical course of patients with PBT was significantly worse. A dose-response relationship was evident but the types of blood products did not influence the outcome. Cox regression analysis adjusting for potentially confounding prognostic factors revealed that the clinical course was not altered by perioperative blood transfusion itself. These observations do not support the idea of adverse effects of perioperative blood transfusion on outcome of patients undergoing 'curative' resection for gastric cancer.