It is not well known if the improvement in operative mortality after surgery for gastric cancer reported in hospital series can be extrapolated to the whole population. The aim of this study was to determine trends in operative mortality over a 20-year period in a nonselected community-based series of patients. A database of 648 patients with gastric cancer resected with curative intent between 1976 and 1995 in a region with a half-million population was divided into two periods: 1976-1983 and 1984-1995. Nonconditional logistic regression was performed to estimate the independent effects of the studied factors. Operative mortality was higher during the 1976-1983 period than during the 1984-1995 period (17.1% vs. 7.1%; p < 0.0001). When comparing the two study periods, operative mortality decreased dramatically from 26.2% to 10.0% in patients over age 70, from 31.8% to 7.9% after total gastrectomy, and from 30.7% to 6.3% after proximal esophagogastrectomy. Operative mortality after total gastrectomy was nearly the same as that after distal gastrectomy (7.9% vs 5.9%) during the second study period. During the first study period, operative mortality was independently associated with age at diagnosis, type of gastrectomy, and to a lesser degree stage at diagnosis; during the second study period, only age and stage at diagnosis were associated with the risk of operative mortality. This study indicates that in this well defined population operative mortality after curative resection for gastric cancer has decreased during the last 20 years. The results should encourage aggressive management of patients with gastric cancer, even in patients over 70 years of age.