This is a retrospective review of 328 consecutive patients with histologically confirmed gastric adenocarcinoma diagnosed in one centre between 1974 and 1984. Of these patients, 128 (39 per cent) had a curative resection, 32 (9.8 per cent) had a palliative resection, 33 (10.0 per cent) had a gastro-enterostomy, 26 (7.9 per cent) had a Celestin tube inserted, 58 (17.7 per cent) had a laparotomy alone, and 51 (15.5 per cent) had no surgical procedure. The 5 year survival was 11 per cent but all long term survivors were patients who had a curative resection. Using multivariate analysis the best predictor of survival after curative resection was the presence or absence of serosal involvement (P = 0.0004). Patients with a long history of presenting symptoms (greater than 6 months) survived longer than those with a short history (P = 0.001). The impact of chemotherapy on the survival of 202 patients with advanced gastric cancer was analysed by multivariate analysis. The median survival of the 50 patients receiving combination chemotherapy was better than that of the 152 who did not (median survivals 160 versus 71 days; P less than 0.001). When deaths occurring within 14 days of diagnosis were excluded, the significance value dropped to P = 0.02. Comparison of the groups treated between 1974 and 1979, when 8 per cent of 92 patients received chemotherapy, with 1980-1984, when 45 per cent of 110 patients received chemotherapy, showed no significant difference in survival.