To evaluate the effect of lymph node dissection on gastric cancer patients operated upon with curative intent, we are carrying out a multicenter, prospective, randomized, controlled study in the Netherlands. The trial compares conventional gastrectomy to gastrectomy with extended lymph node dissection. In the first four months, a Japanese supervisor attended all the extended surgery and instructed many Dutch surgeons, including the eight consulting surgeons; since then, all extended gastrectomies have been attended by one of the consulting surgeons. The study coordinator attended all conventional cases. This assured that the quality of the extended surgery was as good as the Japanese standard, of which excellent results have been reported. To achieve this quality control, randomization before surgery was obligatory for practical reasons. Curability assessment at laparotomy, however, is done quite objectively with histological proof, except for the judgement of irresectability. Although this has resulted in many non-curative cases being randomized but subsequently not given the allocated surgery, the sample size should be sufficient to allow analysis according to randomization or the initial "intention to treat." This is the first protocol for a multicenter trial in surgical oncology to have such excellent surgical quality control and to assure a quality as high as that in the original report with uniformity in the level of technique. In studies comparing surgical techniques, it is vital that attention should be given to surgical quality control, otherwise survival rates may show little improvement and fail to make any impact on surgical practice.