The contention that the R2 radical gastrectomy for localized and potentially curable gastric carcinoma may be superior to gastrectomy without lymphadenectomy (R1) was assessed by randomized trial. Five years after commencement 403 patients have been evaluated at surgery and only 43 (11 per cent) found eligible (S0-2, P0, H0, N0-1), 22 of whom underwent R1 and 21 R2 gastrectomy. Seven patients had final histological stages in excess of the protocol. The R2 group had a longer operating time (P less than 0.005), a greater blood transfusion requirement (P less than 0.005), a longer hospital stay (0.05 greater than P greater than 0.025) and required reoperation in four cases. There were no postoperative deaths. Four patients have died from the disease in the R1 group and five in the R2 group, there being no difference in the probability of survival at a median follow-up of 3.1 years. The small proportion of patients suitable for radical R2 surgery, the high associated morbidity and the fact that survival advantage has yet to be proven in trial suggest that this procedure should not yet be performed outside of controlled clinical trials.