Of the various pathology classifications for gastric carcinoma, those of Lauren and of Ming have been claimed to be of prognostic value. We therefore reviewed the charts of 101 patients (mean age 64 +/- 11 years) treated for gastric carcinoma between 1960 and 1983 at the Institut Curie. We excluded the following patients: 1) those with distant metastases; 2) those who were not operated on; 3) those who underwent laparotomy only, and 4) those who had a palliative procedure. The type of surgery performed was partial gastrectomy in 73 cases and total gastrectomy in 28. Pathology was reviewed according to the WHO, Lauren, and Ming classifications. Were taken into consideration: 1) whether the surgeon believed that removal of tumor was grossly complete or not, 2) the degree of parietal involvement according the pTNM classification system, and 3) the number of positive lymph nodes. Survival was analyzed by the Kaplan Meier method after exclusion of all postoperative deaths. Mean overall survival was 30 months, while that for patients undergoing grossly complete removal of tumor was 34 months. In univariate analysis, no correlation was found between survival and pathology according to Lauren's or Ming's classifications, sex, parietal involvement, or age. Survival was statistically correlated with the size of the tumor (p = 0.015), the gross completeness of surgery (p = 0.008), the type of surgery performed (prognosis was better after partial gastrectomy than after total gastrectomy) and the number of positive lymph nodes (p = 0.0001).(ABSTRACT TRUNCATED AT 250 WORDS)