Early gastric cancer (EGC) is defined as a lesion in which the depth of invasion is limited to the mucosa, submucosa, or both regardless of lymph node status; moreover, it has an excellent prognosis, with a 5-year survival rate of more than 90%. We aimed to determine the prognostic factors for EGC in a large Western series. Over a 10-year period from January 1979 to December 1988 a series of 332 patients (mean age 64 years) with EGC were operated on in 23 centers (two French Associations for Surgical Research). The clinicopathologic data retrospectively and screened it for prognostic effect. The mean follow-up for the 332 EGC patients was 80 months. Postoperative mortality was 4% among 243 partial and 89 total gastric resections. The overall 5- and 7-year survival rates were 82% and 72%, respectively. The cumulative 5- and 7-year survival rates (mean follow-up 80 months) were 92.0% and 87.5%, respectively, excluding both operative and unrelated mortality. There was no significant difference in survival between partial and total gastric resection for lesions located in the lower third of the stomach ( p > 0.6). When survival data (excluding postoperative deaths) were analyzed using univariate analysis and Cox's proportional hazards model, lymphatic involvement ( p = 0.01), the site of the tumor in the upper two-thirds of the stomach ( p = 0.02), and submucosal lesions ( p = 0.049) showed a significant effect on predicting a poor prognosis. These results suggest that because of its prognostic value lymphadenectomy should be performed in addition to gastric resection for adequate classification of EGCs. Follow-up might be required only for patients with at least one poor prognostic factor.