Controversy still exists on the optimal surgical resection for potentially curable gastric cancer. Much better long-term survival has been reported in retrospective/non-randomized studies with D2 resections that involve a radical extended regional lymphadenectomy than with the standard D1 resections. In this paper we report the long-term survival of patients entered into a randomized study, with follow-up to death or 3 years in 96% of patients and a median follow-up of 6.5 years. In this prospective trial D1 resection (removal of regional perigastric nodes) was compared with D2 resection (extended lymphadenectomy to include level 1 and 2 regional nodes). Central randomization followed a staging laparotomy. Out of 737 patients with histologically proven gastric adenocarcinoma registered, 337 patients were ineligible by staging laparotomy because of advanced disease and 400 were randomized. The 5-year survival rates were 35% for D1 resection and 33% for D2 resection (difference -2%, 95% CI = -12%-8%). There was no difference in the overall 5-year survival between the two arms (HR = 1.10, 95% CI 0.87-1.39, where HR > 1 implies a survival benefit to D1 surgery). Survival based on death from gastric cancer as the event was similar in the D1 and D2 groups (HR = 1.05, 95% CI 0.79-1.39) as was recurrence-free survival (HR = 1.03, 95% CI 0.82-1.29). In a multivariate analysis, clinical stages II and III, old age, male sex and removal of spleen and pancreas were independently associated with poor survival. These findings indicate that the classical Japanese D2 resection offers no survival advantage over D1 surgery. However, the possibility that D2 resection without pancreatico-splenectomy may be better than standard D1 resection cannot be dismissed by the results of this trial.