A randomized trial of surgical adjuvant chemoimmunotherapy was conducted in patients who had undergone palliative gastrectomy for previously untreated advanced stomach cancer. First, all patients received the same induction chemoimmunotherapy with MFC (mitomycin C, 5-fluorouracil, and cytosine arabinoside) plus OK-432 for 6 weeks after surgery. The patients were then randomized to receive either chemoimmunotherapy with MFC plus OK-432 (group A) or immunotherapy with OK-432 alone (group B) for maintenance. The survival rate of patients was significantly higher in group B (44 cases) than in group A (39 cases) during the first 9 months after the start of induction therapy (P less than 0.05). A further division of patients in terms of carcinoma histology revealed a difference in survival rate only in patients with an undifferentiated histology (poorly differentiated adenocarcinoma and signet-ring cell carcinoma), and not in those with a differentiated histology (papillary, tubular, and mucinous adenocarcinomas). These results indicate that simple immunotherapy with Ok-432 is better for maintenance than chemoimmunotherapy involving MFC, particularly in patients with undifferentiated gastric carcinomas.