Although lymph node metastasis is a major prognostic factor in gastric cancer, the optimal extent of lymph node dissection still remains a subject of debate. The influence of extended D2 lymphadenectomy on morbidity and long-term survival is controversial. Reports from many Japanese and some Western institutions show similar morbidity and mortality rates for both limited D1 and extended D2 resections. However, the four available randomised trials show a significant increase in operative morbidity and mortality after a D2 resection. The authors of these trials believe that distal pancreaticosplenectomy is responsible for this increased morbidity and mortality and not the lymphadenectomy itself. Retrospective and prospective non-randomised studies show superior stage (II/IIIA) specific survival rates after D2 resections. However, these studies did not eliminate stage migration and randomised trials failed to show any survival advantage in favour of the D2 resection. Current data suggest that D2 resection is beneficial to the subgroup of patients with N1 or N2 disease undergoing potentially curative resection. However, Western studies that support D2 resection, fail to show any survival advantage for D2 resection in N2 patients, reporting a benefit only to N0 or N1 patients. In contrast, Japanese series report a large number of N2 long-term survivors. The question as to the possible beneficial effect of extended lymphadenectomy in gastric cancer is difficult and complex. D2 resection increases the potentially curative resection rate, at least in N2 patients, achieves a better locoregional tumour control and provides the only chance for cure among N2 patients since adjuvant treatment in gastric carcinoma has not yet been proved effective. However, all randomised comparisons warn of an increased risk after D2 resection. By avoiding pancreaticosplenectomy, however, the morbidity can be within acceptable limits. D2 gastrectomy seems to be the most attractive procedure in the surgical management of gastric cancer.