This study investigated whether an endoscopic surveillance program for patients with "benign" gastric ulcers and gastric cancer leads to early detection of neoplasms and improves survival. The clinical course of all patients diagnosed between 1977 and 1986 as having either gastric ulcers or gastric cancer was followed for a minimum of 3 years. Of 597 patients with initially benign gastric ulcers, 452 (76%) returned for the recommended endoscopic follow-up examinations. In eight patients (1.8%), repeated biopsies disclosed malignant neoplasms; four of these patients (0.9%) had become asymptomatic. Survival curves were nearly identical in patients who complied and those who did not. Of 241 patients with gastric cancer, 72 underwent partial gastric resection with curative intent and survived the first year. Resectable cancer was detected in 5 of 48 patients who complied (10%); none of these patients died of cancer. However, 5-year actuarial survival rates were similar between the patients who complied and those who did not. Although endoscopic surveillance may detect resectable cancer in selected patients, it remains to be shown that such a strategy improves survival.